# BioBricks: full content

## Research Synthesis Price Index (updated 2026-08-24)

| Vendor | EOR price | Coverage claim | Status | Source | Checked |
| --- | --- | --- | --- | --- |
| Twist Bioscience |  |  |  |  |  |
| GenScript |  |  |  |  |  |
| Bio Basic |  |  |  |  |  |
| Quintara Biosciences |  |  |  |  |  |

Methodology: Every rendered figure is a price the vendor publishes on its own page, re-fetched and matched verbatim (the quote and its retrieval date travel with the cell), then hash-pinned; a vendor that publishes no price is listed with that fact, never an estimate. Figures are advertised ENTRY prices -- the vendor's own 'from' number -- so sequence complexity, construct length, turnaround and delivery format all move the quoted total, and these compare starting points rather than final invoices.

## Gene synthesis pricing

Gene synthesis is quoted per base pair, and the per-base headline is the number vendors compete on hardest: it is printed on the front of the service page, it moves with promotional cycles, and it is almost never what you finally pay. The table below shows what each vendor advertises as its entry rate, quoted verbatim from its own page on the date shown, so you can see the spread before you send a sequence to anyone for a real quote.

1. **The per-base entry rate**: Every vendor advertises a 'from' price per base pair, and that figure assumes the easy case: a sequence of ordinary length with no difficult structure, in the vendor's standard vector, on standard turnaround. It is a real price for real orders, but it is the floor of a range rather than a quote.
2. **Sequence complexity re-tiers the rate**: High GC content, long repeats, homopolymer runs and strong secondary structure all push a construct into a higher complexity tier, and the same vendor will quote a materially higher per-base rate for it. Vendors that publish more than one rate are showing you the tier structure directly; ones that publish a single rate still apply it.
3. **Cloning, vector and prep are separate lines**: A synthesised fragment is not a usable plasmid. Cloning into an expression vector, the vector itself, and the plasmid prep that gives you working DNA are priced separately by most vendors, which is why a flat-rate-per-construct price and a per-base price are not comparable until you add the same lines to both.
4. **Turnaround is a price lever, not a service level**: Express turnaround is sold as an uplift on the same construct. When you are comparing two vendors whose per-base rates look close, check what each charges to hit the date you actually need, because that surcharge routinely exceeds the per-base difference on a short construct.

**How much does gene synthesis cost per base pair?**

The vendors in this index that publish a rate advertise entry prices spanning the range shown in the table above, each quoted verbatim from the vendor's own page with the date we checked. Those are entry rates for standard-complexity sequences; a difficult construct is quoted higher by the same vendor.

**Why is my quote higher than the advertised price?**

Almost always sequence complexity, or lines the headline excludes. High GC, repeats and secondary structure move a construct into a higher tier, and cloning, the vector and plasmid prep are separately priced by most vendors. Ask for the quote itemised and the gap is usually explained on the first line.

**Is a flat rate per construct cheaper than a per-base price?**

It depends entirely on construct length. A flat rate wins on longer constructs and on anything where you would have paid separately for cloning and prep; a per-base rate wins on short fragments delivered bare. This index publishes both wherever a vendor advertises both, for exactly that reason.

**Do these prices include cloning into my vector?**

Only where the vendor says so. One flat-rate price in this index explicitly includes cloning and plasmid preparation, and its source quote says as much; the per-base rates generally do not. The quote beside each figure is the vendor's own wording, so you can check what is covered.

## Custom oligo synthesis

Custom oligos are the highest-volume line most labs buy, and the pricing is deceptively layered: a per-base rate, a synthesis scale, a purification grade and any modification are four separate multipliers on the same order. A plain desalted 20-mer at the smallest scale and an HPLC-purified modified oligo at a larger scale are the same product line and differ in price by more than an order of magnitude.

1. **Length, charged per base**: The base rate is per nucleotide, so a 40-mer costs roughly twice a 20-mer before anything else applies. Vendors publish this rate at their standard scale, and it is the only one of the four levers that scales linearly and predictably.
2. **Synthesis scale**: Scale is the amount of material made, quoted in nanomoles or micromoles. Ordering a larger scale than an assay needs is the most common way labs overspend on oligos; ordering too small means re-ordering, which costs the lead time as well as the money.
3. **Purification grade**: Desalted is the default and is fine for routine PCR primers. HPLC or PAGE purification is charged as an uplift and is genuinely needed for longer oligos, for anything going into a sequencing library, and for applications where a truncated failure sequence would confound the result.
4. **Modifications**: Fluorophores, quenchers, phosphorylation, biotin and locked nucleic acids are each priced per modification, and dual-labeled probes carry the largest uplifts in the whole category. A modified probe order is not comparable to a plain primer order on the per-base rate alone.

**What does a custom oligo cost?**

Per base, at the vendor's standard scale, with desalting. The table above shows what the vendors in this index advertise, quoted from their own pages. A modified or HPLC-purified oligo costs substantially more, and the uplift is priced separately by every supplier.

**What scale should I order, and when is oligo pool synthesis right?**

The smallest that covers your planned reactions with a margin. For routine PCR primers that is usually the smallest scale offered; ordering larger 'to be safe' is the most common oligo overspend, because primers are cheap to re-order and do not improve with age.

**Do I need HPLC oligo purification?**

Not for routine PCR primers, where desalted is standard and adequate. Yes for long oligos, dual-labeled probes, and anything entering a sequencing library, where truncated failure sequences from the synthesis would otherwise carry into the result.

**Why do modified oligos cost so much more?**

Each modification is an extra coupling step with its own reagent, and dual-labeled probes need two, plus a purification the plain oligo would not require. Vendors price each modification as a separate line, so a probe order is best compared modification by modification rather than on the per-base rate.

## Plasmid DNA preparation

Outsourced plasmid prep is bought for two quite different reasons: to save bench time on routine constructs, and to get a grade of DNA a normal lab cannot reliably produce. The price gap between those two jobs is enormous, and it is set almost entirely by two variables that appear on every quote: how much DNA you need, and how clean it has to be.

1. **Scale, in milligrams of DNA**: Quotes start from the yield you need, not from the culture volume. A few hundred micrograms for cloning and transfection screening is a different product from tens of milligrams for a transfection campaign, and vendors publish separate price bands rather than a linear rate.
2. **Grade: research, transfection or GMP**: Research grade is standard column-purified DNA. Transfection grade adds endotoxin removal, which matters because endotoxin kills primary cells and skews every downstream readout. GMP grade adds a documented quality system and is priced in a different bracket entirely.
3. **Endotoxin specification**: Where endotoxin is specified, it is specified as a limit in endotoxin units per milligram, and the tighter the limit the higher the price. This is the single line that most often explains why two quotes for 'the same' prep differ by several multiples.
4. **QC included, and QC you have to ask for**: Most preps include concentration, purity ratios and a restriction digest check as standard. Full-plasmid sequencing confirmation is increasingly cheap and is worth adding explicitly, because a prep of the wrong construct is the failure mode that costs the most downstream time.

**What do plasmid DNA manufacturers charge?**

It is banded by yield and grade rather than quoted per milligram. The vendors in this index that publish a figure are shown above with the quote and date; a research-grade prep at modest scale and an endotoxin-specified transfection-grade prep at high scale sit in different brackets entirely.

**Do I need a transfection grade plasmid DNA vector or research grade?**

Research grade is fine for cloning, bacterial work and robust immortalised cell lines. Transfection grade with endotoxin removal is worth the uplift for primary cells, stem cells, immune cells and anything going into an animal, where endotoxin measurably changes the result.

**What endotoxin limit should I ask for?**

Ask for the limit your downstream assay tolerates, and require it as a specification on the certificate rather than as a description. A vendor that will state a numeric limit per milligram is committing to it; one that describes DNA as 'low endotoxin' without a number is not.

**Should I pay a DNA cloning service for sequencing confirmation?**

Usually yes. Whole-plasmid sequencing has fallen to a small fraction of the prep cost, and it is the only check that catches a rearrangement or a wrong clone before the DNA reaches an experiment. It is cheap insurance against the most expensive failure mode.

## Custom synthesis services

"Custom synthesis" covers a wide span of work: nucleic acids, peptides, small molecules and proteins, all made to a customer specification rather than sold from a catalog. The pricing models differ by material, but the buying problem is the same one every time, and it is not really about price: it is about writing a specification precise enough that two vendors quote the same job.

1. **The specification comes before the quote**: A custom quote is only as good as the specification behind it. Identity, quantity, purity, the analytical evidence you require and the delivery format are the five things that must be fixed before any number is meaningful, because each of them is a lever the vendor prices.
2. **Purity and the analysis that proves it**: Purity is a specification, not an adjective, and the method that demonstrates it is a separate decision. Ask which analyses are included, what the acceptance criterion is, and whether you receive the raw data or a summary certificate.
3. **Scale changes the route, not just the number**: Beyond a certain quantity a vendor may change synthetic route entirely, which changes cost per unit, lead time and impurity profile together. This is why per-unit prices quoted at research scale rarely extrapolate to a larger order, and why you should quote the scale you will actually need.
4. **Risk sits somewhere, and it is priced**: Difficult custom work either carries a guaranteed deliverable at a higher price, or a lower price with the risk of failure on you. Both are legitimate; what matters is knowing which you bought. Ask explicitly what happens, and what you pay, if the synthesis does not reach specification.

**What is a custom synthesis service?**

A supplier making a specified material to order rather than selling it from a catalog: nucleic acids, peptides, small molecules or proteins. It is bought when the material you need is not stocked, or when the quantity, purity or format you need differs from the catalog product.

**How is custom synthesis priced?**

By material class. Nucleic acids are priced per base at a stated scale; peptides per residue with purity and modification uplifts; small molecules per project against a route. The vendors in this index that publish figures are shown above with their own wording and the date.

**What should a quote include?**

Identity, quantity, purity specification, the analytical evidence supplied, delivery format and lead time, plus what happens if the synthesis fails to meet specification. A quote missing the last item is not comparable to one that includes it.

**How do I compare two custom chemical synthesis quotes fairly?**

Fix the specification first and send the identical text to both. Differences that remain are real vendor differences; differences that come from each vendor assuming a different purity or scale are an artefact of a loose brief, and they are the usual reason quotes look incomparable.

## RNA sequencing services

An RNA sequencing quote is built from three things that are often bundled into one number: library preparation, sequencing depth, and the analysis you get back. Vendors bundle them differently, so a per-sample price is only comparable once you know how many reads it buys, which library chemistry it assumes, and whether a differential expression analysis is included or sold separately.

1. **Library preparation chemistry**: Poly-A selection is the standard and cheapest choice for intact eukaryotic mRNA. Ribosomal depletion costs more and is required for degraded samples, bacterial RNA and any work where non-polyadenylated transcripts matter. The chemistry is chosen by your sample, not by your budget, and it changes the per-sample price before depth is considered.
2. **Depth, in reads per sample**: Depth is the main dial on cost. Standard differential expression on a well-annotated genome needs far fewer reads than isoform-level work or low-expression transcript discovery, and paying for depth an experiment cannot use is the most common way an RNA-seq budget is wasted.
3. **Read length and pairing**: Single-end short reads are adequate for counting transcripts against a reference. Paired-end and longer reads cost more and earn it only where isoforms, fusions or novel transcripts are the point of the experiment. Match this to the question, not to the best specification offered.
4. **Analysis: included, extra, or yours**: Some quotes stop at demultiplexed FASTQ files, others include alignment, counts and a differential expression report. That difference is easily the largest uncontrolled variable between two per-sample prices, and it is the first thing to normalise when comparing.

**How much does RNA sequencing cost per sample?**

It depends on library chemistry, depth and whether analysis is included. The vendors in this index that publish a per-sample figure are shown above, quoted verbatim with the date; treat a published per-sample price as an entry point and confirm what depth and analysis it assumes.

**How many reads do I need?**

Standard differential expression against a well-annotated reference needs far fewer reads than isoform-level or discovery work. Set depth from the question you are answering, and put spare budget into biological replicates before extra reads.

**NGS prep for RNA: how much RNA prep does the library need?**

Poly-A for intact eukaryotic mRNA, which is most experiments and is cheaper. Ribosomal depletion for degraded RNA, bacterial samples, and any work where non-polyadenylated transcripts are part of the question.

**Is analysis included in the price?**

Sometimes, and it is the biggest hidden difference between quotes. Ask whether the price ends at FASTQ files or includes alignment, counts and a differential expression report, and normalise for that before comparing per-sample figures.

**What should a sequencing analysis deliverable include?**

Quality metrics, the reference and annotation versions, the statistical model, the pipeline as code with pinned versions, and the intermediate files, not only a table of differentially expressed genes. Re-analysis is the normal fate of these datasets.

**Panel, exome or genome, and what the whole genome sequencing cost decides?**

A panel for a defined set of regions at depth, an exome for coding variants at moderate depth, a genome for structural and non-coding questions. A panel cannot report anything outside its regions.

**Why is a whole genome sequencing price meaningless without depth?**

Because depth is most of the cost and most of what determines what can be detected. Ask for depth or yield per sample, the policy if quality control fails, and how long raw data are retained.

## Custom siRNA synthesis

Custom siRNA sits between routine oligo synthesis and specialist nucleic acid manufacture. It is priced like an oligo ,  per synthesis, at a stated scale, with purification and modification uplifts ,  but the purity requirements are higher and the failure mode is subtler: an under-purified duplex does not fail visibly, it produces a knockdown result you cannot fully trust.

1. **Scale, in nanomoles of duplex**: siRNA is quoted per synthesis at a nominal scale, and the delivered quantity of annealed duplex is what you should compare, not the scale of the individual strands. Screening work needs little material; a sustained knockdown campaign across many plates needs materially more and is priced in a different band.
2. **Purification, which is not optional here**: Unlike routine PCR primers, siRNA generally warrants HPLC purification. Truncated strands compete in the duplex and dilute the effective concentration, so a nominally identical order at lower purity delivers less real knockdown, which reads as biology rather than as a reagent problem.
3. **Chemical modifications**: 2'-O-methyl and phosphorothioate modifications improve nuclease stability and reduce off-target effects, and each is priced per modified position. For cell culture screening they are often unnecessary; for anything with a longer time course or an in vivo component they change the experiment enough to be worth the uplift.
4. **Annealing and QC**: Ask whether the quoted price includes annealing the two strands and confirming the duplex, or delivers single strands you anneal yourself. Both are sold, they are not the same product, and the difference is a common reason two siRNA quotes look far apart.

**What does custom siRNA synthesis cost?**

It is quoted per synthesis at a stated scale, with purification and modification uplifts on top. The vendors in this index that publish figures are shown above with their own wording and the date we checked; compare on delivered duplex quantity rather than on strand scale.

**Do I need HPLC purification for siRNA?**

In most cases yes. Truncated failure strands compete in the duplex and reduce effective knockdown, and because the result still looks like biology rather than a reagent fault, under-purified siRNA is unusually expensive in wasted experiments.

**Are chemical modifications worth the uplift?**

For short cell-culture screens, often not. For longer time courses, serum-containing conditions and any in vivo work, 2'-O-methyl and phosphorothioate modifications improve stability and reduce off-target effects enough to change what the experiment can show.

**How many sequences should I order per target?**

At least two independent duplexes, ordered at screening scale. Showing the same phenotype with independent sequences is the standard defense against off-target artefacts, and it is cheaper to buy several small syntheses than to scale up the wrong one.

## LIMS selection

A laboratory information management system is the record of what came into the lab, what was done to it, and what was reported out. Buying one is less a software decision than a decision about how your lab will be audited for the next decade, because the system you choose becomes the place your results live. This page sets out how the category is priced, what separates the vendors, and what to settle before you sit through a demonstration.

1. **Named seats, not headcount**: Almost every vendor prices per named user or per concurrent session, and the two are very different for a lab where thirty people touch the system but four do so at once. Establish which model is quoted before you compare anything, because the same lab can be two very different numbers under the two schemes.
2. **Configuration is the hidden line**: The license is rarely the largest number in year one. Workflow configuration, instrument interfacing, report templates, migration of legacy results and validation documentation are quoted as a project, and that project is where most of the first-year spend sits. Ask for it itemised against a named scope rather than a day-rate estimate.
3. **Instrument integration is priced per interface**: Connecting a balance, a plate reader or a chromatography data system is normally a per-instrument charge, and the price depends on whether the vendor already has a driver for that make and model. Bring your instrument list to the first call, because a lab with unusual or elderly kit discovers this line late and it is not small.
4. **Validation and qualification for regulated work**: If your results go into a regulated submission, the system needs installation, operational and performance qualification, and the documentation has to be maintained through upgrades. Some vendors sell a validation pack; others expect you to write it. That difference is worth more than a modest license discount.
5. **The exit price nobody quotes**: Ask what it costs to get your data out in a usable form, and in what format, on the day you leave. A vendor that answers this clearly is telling you something real about how the contract will feel in year five.

**What is LIMS?**

A laboratory information management system: software that registers samples, tracks them through testing, stores the results against the method and the analyst, and produces the report and the audit trail. It is the lab's system of record rather than a document store.

**How do LIMS vendors charge?**

Usually a per-user license, annual or perpetual with maintenance, plus a separate implementation project covering configuration, instrument interfaces, data migration and training. The implementation commonly costs more than the first year of license.

**Do I need a LIMS or an ELN?**

A LIMS is built around samples and repeatable tests; an electronic lab notebook is built around experiments and narrative. A quality-control or testing lab usually needs the first, a discovery lab the second, and many labs eventually run both with a defined boundary between them.

**How long does a LIMS implementation take?**

It depends almost entirely on how many instrument interfaces and custom workflows are in scope, not on the size of the lab. Ask each vendor for a reference customer with a comparable instrument list and ask that customer how long it really took.

**How should I compare systems?**

Against requirements you wrote from your own workflows, marked essential or desirable, before any demonstration. Every product demonstrates well against its own script.

**Why is the implementation more expensive than the licence?**

Because configuration, data migration, instrument interfaces, validation and training are the real work. Interfaces in particular are often priced per instrument type and are where a modest estate becomes an expensive project.

## Laboratory information management systems

The phrase laboratory information management system describes a scope rather than a product, and the scope is the first thing two vendors will disagree about. One will include stability scheduling and environmental monitoring; another will call both of those separate modules with separate prices. This page explains what the system is normally expected to own, where the boundaries with other lab software fall, and how to make two quotes describe the same thing.

1. **Sample registration and chain of custody**: Every sample enters once, with its source, its container, its condition on receipt and its storage location, and every subsequent movement is recorded against a person and a time. This is the part an auditor will test first and the part that is most painful to retrofit.
2. **Test assignment against a method**: A sample is assigned tests, each bound to a version of a method with its own specification limits and its own trained-analyst list. When a method is revised, the system has to keep old results tied to the version they were produced under rather than silently re-pointing them.
3. **Result capture and review**: Results arrive by instrument interface or by keyed entry, are checked against specification, and pass through a review step that a second person performs. Systems differ sharply in how much of this they enforce and how much they merely record.
4. **Reporting and release**: The certificate of analysis, the trend report and the export to whatever the customer or the regulator expects are produced from the same record, with the release decision recorded against a named person. A system that produces reports by exporting to a spreadsheet has moved the record out of itself.

**Is a LIMS laboratory management system the same as a laboratory information management system?**

Yes. LIMS is simply the initialism, and vendors use the two interchangeably in their own material. The long form is more common in tenders and in regulatory documents.

**Can a laboratory information management system replace our spreadsheets?**

For sample tracking, result capture and reporting, yes, and that is the point of buying one. Spreadsheets usually survive for method development and for one-off analyses, which is reasonable as long as nothing released to a customer depends on them.

**Does the system need to be validated?**

If your results support a regulated submission or a regulated product, the installed configuration needs qualifying and the qualification needs maintaining through upgrades. A research lab with no regulatory exposure can take a much lighter approach, but should still record what it configured and why.

## Erlenmeyer flasks

The conical flask is the most ordered item in most laboratories and the one buyers think least about, which is why store rooms fill with flasks nobody can use for the job in hand. The differences that matter are the glass or resin grade, whether the graduations are indicative or measured, the neck and closure standard, and whether the flask is baffled. This page sets out what to put on the order line so the delivery is usable.

1. **Glass grade or resin**: Borosilicate handles thermal shock and direct heating and is the default for anything warm. Polypropylene and polycarbonate are lighter, do not shatter and suit cell culture and buffer work, but each has its own solvent incompatibilities and its own upper temperature. Match the material to the chemistry rather than to the price.
2. **Graduations are indicative, not metrological**: Molded graduations on a conical flask are there to guide, not to measure. If a volume matters, it is measured in a volumetric flask or a cylinder and transferred. Ordering expensive flasks in the belief that the printed marks are accurate is a common and avoidable mistake.
3. **Neck and closure standard**: Narrow and wide mouth are not interchangeable, and ground-glass joints carry a size designation that has to match the stopper or condenser you already own. Culture flasks add vented and non-vented caps, which are a sterility decision rather than a convenience.
4. **Baffled or smooth**: Baffles increase gas transfer for shaking cultures and make the flask much harder to clean and impossible to use for accurate pouring. Buy baffled flasks for culture and smooth flasks for everything else rather than compromising on one stock item.

**What is an Erlenmeyer flask used for?**

Mixing, heating and swirling liquids without spilling, and as a culture vessel on a shaker. The conical body allows vigorous swirling, and the narrow neck takes a stopper or a closure and reduces evaporation.

**Can you measure volume in an Erlenmeyer flask?**

Not accurately. The molded graduations are approximate by design. Use a graduated cylinder or a volumetric flask when the volume is part of the result.

**Glass or plastic Erlenmeyer flask?**

Borosilicate glass for heat, solvents and long service life; polypropylene or polycarbonate where breakage is the main risk or the flask goes into a cell culture workflow. Check the manufacturer's chemical compatibility table rather than assuming.

**What are baffled flasks?**

Indentations molded into the base that break up the swirl and increase oxygen transfer in a shaking culture. They improve growth and make the flask considerably harder to clean.

**Are Erlenmeyer flasks autoclavable?**

Borosilicate yes, repeatedly, if undamaged. Plastics vary by resin and have a limited cycle life stated by the manufacturer; polycarbonate in particular degrades noticeably with repeated cycles.

**Why does my culture stop growing at a low density?**

Usually oxygen transfer. A smooth flask filled too full starves the culture. Fill to a fifth or less of nominal volume and use a baffled flask where the organism is demanding.

## Fume hoods

A fume hood is the only piece of lab equipment whose purchase price is routinely the smaller half of the spend. The enclosure is quoted by the manufacturer; the ductwork, the fan, the make-up air, the controls and the commissioning are quoted by somebody else, and a hood ordered without that second conversation sits in a crate. This page covers the specification decisions and the parts of the project that the equipment quote does not include.

1. **Ducted or filtered**: A ducted hood exhausts outside and needs a fan, a duct route and make-up air. A filtered or ductless hood carries its own carbon or particulate filtration and needs none of that, but it is only suitable for a known and limited chemical inventory, and the filters are a recurring cost with a monitored breakthrough point.
2. **Bypass, restricted bypass or variable air volume**: The airflow scheme decides how much conditioned air the hood throws away when the sash is down. Variable air volume costs more to install and much less to run, which usually decides it for a new building and rarely decides it for a single replacement hood.
3. **Width, depth and interior services**: Hood width is chosen from the work, not the space, and every service dropped into the interior, gas, vacuum, water, electrical outlets, is a separate line and a separate penetration. Decide the interior layout before the order rather than during the installation.
4. **Liner and worktop material**: Epoxy resin is the general-purpose liner; stainless steel suits radionuclide and perchloric work, and polypropylene suits hydrofluoric acid. A liner chosen for price and not for chemistry is replaced early, and replacing a liner means taking the hood out of service.
5. **Commissioning and the containment test**: The hood is not usable until face velocity is set and a containment test has been performed as installed, with the room's air balance as it will actually run. Confirm who performs that test, when, and what happens if the result fails.

**What is the difference between a fume hood and a biosafety cabinet?**

A fume hood protects the user from chemical vapour by drawing air away and exhausting it. A biosafety cabinet protects the user, the sample or both using filtered air, and many recirculate. They are not interchangeable: a fume hood offers no product protection and a cabinet offers limited chemical protection.

**Do ductless fume hoods work?**

For a defined and limited chemical inventory within the filter's capability, yes, and they are far cheaper to install. They are unsuitable for unknown or varied chemistry, for large volumes and for anything the filter medium does not retain, and they need a monitored filter change regime.

**How is a fume hood tested?**

Face velocity across the sash opening is measured at commissioning and periodically afterwards, and a containment test using a tracer is performed as installed. A hood that has never been tested as installed in its own room has not been proven to contain anything.

**What size fume hood should I buy, and when is a benchtop fume hood enough?**

Size it from the largest apparatus you will put in it plus working clearance, then check the room can supply the exhaust. Buying a wider hood than the air handling supports is the most expensive way to discover a building limit.

**Is a second-hand enclosure a sensible buy?**

Only with the extract system it will be connected to. The enclosure is the cheap part, and its containment performance depends on ductwork, fan sizing and commissioning. Budget a containment test as installed, whatever its history.

**Does a capture arm replace an enclosure?**

No. It captures what is released close to it and does not contain a spill or a runaway reaction. It suits a known, localised, low-hazard release where the operator keeps the source within a short distance.

## Laminar flow hoods

A laminar flow hood pushes filtered air across a work surface to keep particles off whatever is on it. That sentence contains the whole safety argument: it protects the product, not the person. Labs that buy one expecting containment are the reason this distinction is worth stating before any specification is discussed. This page covers the flow geometries, what each is good for, and what has to be true of the room around it.

1. **Horizontal flow blows toward the operator**: Air passes through the filter at the back and travels across the work surface toward you. Product protection is excellent and there is no barrier between the work and your face, so it suits media plating, electronics and sterile assembly of non-hazardous material and nothing else.
2. **Vertical flow drops from above**: Air passes down through a filter in the roof of the cabinet onto the work surface. It is friendlier to bulky apparatus, spares the operator a direct stream of anything volatile, and is the usual choice where the work surface is crowded.
3. **Neither contains an aerosol**: A laminar flow hood exhausts into the room. If the work generates an infectious or toxic aerosol, the correct enclosure is a biosafety cabinet, which filters the exhaust and holds an inflow barrier at the opening. This is the single decision on this page that has consequences beyond cost.
4. **Filter integrity is certified, not assumed**: The high efficiency filter is tested in place at installation and periodically afterwards, and the test is what tells you the seal and the medium are intact. A hood with an unknown certification date is producing clean air only in principle.

**Does a laminar flow hood protect the operator?**

No. It protects the work from airborne particles and exhausts into the room. For operator or environmental protection you need a biosafety cabinet or a fume hood, depending on whether the hazard is biological or chemical.

**Horizontal or vertical laminar flow hoods?**

Horizontal gives the cleanest presentation to the work and blows toward the operator. Vertical is better with bulky apparatus and keeps the stream off the operator's face. Choose vertical if there is any doubt.

**How often should the filter be certified?**

At installation, after any move, and on a periodic schedule set by your quality system. The date and the method should be recorded on the cabinet where anybody working at it can see them.

**Is an ISO class 5 laminar flow hood guaranteed to meet that class on delivery?**

No. It is measured in the working zone after installation and depends on the room and on how the operator works. Ask who certifies it, how often, and what happens if it fails.

## Graduated cylinders

A graduated cylinder is the everyday measuring vessel of the laboratory, and the only thing that distinguishes a good one from a cheap one is a tolerance you cannot see. The grade on the side of the cylinder tells you how far the printed marks may be from the truth, and whether the manufacturer will issue a certificate saying so. This page covers the grades, the materials and the reading technique that between them decide whether a volume is a measurement or a guess.

1. **Grade decides the tolerance**: Class A ware is manufactured to the tighter tolerance and can be supplied with a certificate traceable to a national standard. Class B is roughly double the permitted error and is the right buy for making up washes and waste. Buying Class A for everything is money spent on an accuracy nobody records.
2. **Material decides where it can go**: Borosilicate glass reads cleanly, resists solvents and survives autoclaving, but it breaks. Polypropylene and polymethylpentene survive drops and cold work; polypropylene is translucent and harder to read, polymethylpentene is clear and more expensive. Match the choice to whether the cylinder lives on a bench or in a cold room.
3. **Base and bumper guard**: A hexagonal or wide round base and a removable plastic bumper guard are what keep a tall cylinder upright and intact. They cost very little and they are the single most effective way to reduce breakage, which is the real recurring cost of glass volumetric ware.
4. **Single or double graduations**: Double scales let you read from the top for a pour-out measurement and from the bottom for a fill. A lab that decants routinely should specify them deliberately rather than discovering afterwards that every reading needs arithmetic.

**What is the difference between Class A and Class B?**

Class A ware is made to a tighter permitted deviation and can carry a traceable certificate; Class B allows roughly twice the error and normally carries none. The marks look identical, so the grade printed on the cylinder is the only way to tell.

**Which material does graduated cylinder chemistry work need, glass or plastic?**

Glass for solvents, heat and clear reading; plastic where breakage is the main cost or the cylinder goes into a freezer or a field kit. Check solvent compatibility before switching a chemistry workflow to plastic.

**Should I read the top or the bottom of the meniscus?**

The bottom, for water and most aqueous solutions, at eye level. Liquids that do not wet the surface form a meniscus the other way and are read at the top; mercury is the classic example and is rare in modern labs.

**Can a graduated cylinder be autoclaved?**

Borosilicate yes, and polypropylene usually yes within the manufacturer's cycle limit. Repeated autoclaving eventually clouds plastic and can shift graduations printed rather than molded, so check how the scale was applied.

**Which vessel should I prepare a standard in?**

A volumetric flask of the tighter tolerance class, made up to the mark at the calibration temperature. A graduated cylinder is calibrated to deliver and is a bulk measuring vessel, not a standard-making one.

## Pipettes and pipette service

A pipette is the instrument that decides whether an assay is reproducible, and it is the instrument most likely to be out of specification without anyone noticing. Buying one is a short decision; keeping a fleet of them inside tolerance is a service contract that runs for the life of the lab. This page covers both, because labs that treat them as separate purchases routinely buy good pipettes and then let them drift.

1. **Air displacement or positive displacement**: Air displacement handles aqueous liquids and is what most labs mean by a pipette. Positive displacement uses a disposable piston in contact with the sample and is what viscous, volatile, dense or contaminating liquids require. Using the wrong one is a systematic volume error, not a preference.
2. **Channel count and spacing**: Single channels suit method development and awkward vessels; eight and twelve channel heads suit plates and are where most of the throughput gain sits. Adjustable spacing heads bridge tubes and plates and cost considerably more, so buy them for a workflow that genuinely moves between formats.
3. **Ergonomics are a real specification**: Plunger force, tip ejection force and weight decide whether a technician can work a full day without strain, and repetitive strain in a pipetting role is a genuine occupational risk rather than a marketing angle. Have the people who will use the instrument hold it before the order goes in.
4. **Tip fit is a lock-in decision**: Every manufacturer will tell you generic tips fit; the seal at the tip cone is what sets the delivered volume, and a poor fit shows up as imprecision the calibration certificate did not predict. Decide deliberately whether you are buying into that manufacturer's tips for the life of the instrument.
5. **The service contract**: A pipette service should state the calibration standard applied, the number of volumes tested, the number of replicates at each, the acceptance criteria, and whether adjustment is included or charged when a unit fails. A certificate that reports a single volume tells you very little about the instrument.

**How often should a micro pipette be calibrated?**

Set the interval from how heavily the instrument is used and what depends on the result. A regulated or high-throughput instrument is typically serviced far more frequently than an occasionally used one, with a quick internal gravimetric check between external services.

**What does a pipette service include?**

At minimum a clean, a seal and tip cone inspection, a multi-volume multi-replicate gravimetric test against stated limits, adjustment where needed and a certificate reporting the measured values. Confirm whether adjustment and parts are included or billed separately.

**Can I use generic tips?**

Often, but the tip cone seal sets the delivered volume, so any change of tip should be re-verified gravimetrically on the instruments concerned. Labs under an accredited scheme generally validate the specific tip and instrument pairing.

**When do I need a positive displacement pipette?**

For viscous, volatile, dense, foaming or highly contaminating liquids, where the air cushion in a conventional pipette introduces a systematic error or lets vapour reach the instrument.

**Is an electronic multichannel pipette worth the extra cost?**

Where the work is repetitive, where several people share a protocol or where operator variation is visible in the data, yes, because it holds speed and stopping point constant. For occasional use a manual instrument is cheaper to own and easier to service.

**Why does my pipette deliver the wrong volume with solvent?**

Because the usual mechanism relies on a cushion of air that expands with vapour pressure and changes with density and viscosity. For volatile or viscous liquids use a positive displacement instrument, where a disposable piston contacts the liquid directly.

## Serological pipettes

Serological pipettes are bought by the case and thought about once a year, which is exactly why the wrong choice is expensive: a size that does not match the culture vessels in use wastes both plastic and time on every single transfer. This page covers the size mix a culture lab actually needs, what sterility and packaging claims mean, and the specification details that decide whether a pipette is usable in a cabinet.

1. **The size mix, not the size**: Buy the mix that matches your vessels and your media volumes rather than standardising on one size. A culture lab feeding flasks and plates typically burns through the small and mid sizes and barely touches the largest, and an order that ignores that ends with shelves of unusable stock.
2. **Plugged or unplugged**: A cotton or hydrophobic plug at the top stops liquid and aerosol reaching the pipette controller and stops contamination traveling the other way. For any sterile culture work the plug is not optional; unplugged pipettes belong in non-sterile chemistry.
3. **Individual or bulk wrap**: Individually wrapped pipettes protect sterility until the moment of use and generate more waste and more handling. Bulk-wrapped bags are cheaper per unit and are only appropriate where the whole bag will be used inside a cabinet in a short period.
4. **Graduation direction and negative graduation**: Many serological pipettes carry a graduation beyond the nominal volume so the full nominal can be delivered. Confirm the direction of the scale and whether it reads to the tip, because mixing two scale conventions in one lab is a quiet source of transfer error.

**What is a serological pipette used for?**

Transferring measured volumes of media, buffers and cell suspensions, usually with a powered or bulb controller, in cell culture and general liquid handling. The graduated barrel makes it a measuring as well as a transfer device.

**Do serological pipettes need a plug?**

For sterile culture work, yes. The plug stops liquid and aerosol reaching the controller and protects the sample from anything inside it. Unplugged pipettes are for non-sterile use.

**Are serological pipettes reusable?**

Glass ones historically were, and are still used in some labs with a wash and autoclave cycle. Modern plastic serological pipettes are single use and are not designed to survive reprocessing.

**What tolerance do serological pipettes hold?**

They are graduated transfer ware rather than certified volumetric ware. Where a volume is part of a reported result, measure it with ware of a stated Class A or Class B grade, or with a calibrated piston pipette.

**Which size should I use for a small volume?**

The smallest instrument that holds it. Accuracy is best in the upper part of each size's range, so measuring a few millilitres with a large instrument is the most common source of error in this kind of transfer.

## Analytical balances

Readability is the number in the advertisement and repeatability is the number that decides whether your weighings are usable. An analytical balance quoted on its finest displayed division, installed on the wrong bench in a room with a door that opens, will not deliver that division and no service visit will fix it. This page separates the specification that matters from the specification that sells, and covers the siting and calibration regime that make the instrument worth its price.

1. **Readability against repeatability**: Readability is the smallest increment the display shows. Repeatability is the spread you actually get weighing the same object repeatedly in your own room. Ask the supplier for the repeatability specification and treat the readability as a display characteristic rather than a measurement capability.
2. **Capacity, and why bigger is worse**: A balance with a larger capacity at the same readability is a compromise, and semi-micro and micro balances trade capacity for resolution deliberately. Size the capacity to the heaviest thing you will weigh including its container, and no further.
3. **Internal or external calibration**: An internal motorised weight lets the balance adjust itself on a temperature change or a schedule, which matters in a room that is not tightly controlled. External calibration with certified weights is cheaper to buy and depends on somebody actually doing it.
4. **Siting decides the result**: Put the balance on a heavy, isolated bench away from doors, walkways, air conditioning outlets and anything that vibrates, and let it stabilise after moving. A fine balance in a poor location performs worse than a coarse balance in a good one.
5. **Static and draughts**: Static on plastic weighing vessels and low humidity together produce drifting readings that look like instrument faults. An ioniser, antistatic vessels or a controlled humidity range fix them, and none of that appears in the purchase specification.

**What is the difference between an analytical and a precision balance?**

An analytical balance reads to a finer division and is enclosed in a draught shield; a precision balance offers more capacity at a coarser division and usually no shield. The choice follows the smallest mass you must weigh reliably.

**How often should an analytical balance be calibrated?**

Adjust and verify on a schedule set by use and by what depends on the result, and verify with certified weights before critical work. Many labs verify daily with a single check weight and have the instrument serviced and adjusted externally on a fixed cycle.

**Why does my balance drift?**

Most often the environment rather than the instrument: draughts, vibration, a temperature change, static on the vessel, or an object not in thermal equilibrium with the room. Rule those out before calling a service engineer.

**What is minimum sample weight?**

The smallest mass the balance can weigh while still meeting a stated measurement uncertainty. Weighing below it produces a number on the display that does not support the accuracy the method assumes.

**What limits analytical balance accuracy in a real room?**

Almost always the environment: draughts, vibration, temperature gradients or static. Site it on a heavy isolated bench away from doors, equilibrate samples, use a static eliminator with powders and keep the shield closed.

**Is internal calibration enough?**

It adjusts the balance against its own reference and proves nothing externally. Verify with certified weights at a stated interval and record it; that chain is what makes a weighing defensible.

## Centrifuges

A centrifuge is bought once and lived with for a long time, and the instrument is the cheaper half of the decision. The rotors decide what protocols you can run, the rotors are what wear out, and the rotors are what turn a routine purchase into a safety question when human or infectious material is involved. This page covers how to specify the machine from the protocols rather than the catalog, and what to ask about containment and service before ordering.

1. **Work in relative centrifugal force, not rpm**: Protocols are written in relative centrifugal force because it is the physical quantity that separates material. The same rotational speed in two rotors of different radius gives very different force, so a machine specified on its top speed alone may not reach the force a method requires.
2. **Fixed angle, swinging bucket or continuous**: Fixed angle rotors pellet quickly and give a sloped pellet; swinging buckets give a flat pellet and suit gradients and plates; each has its own maximum force and its own tube set. Decide which protocols must run before choosing the machine, because the rotor range available is the real constraint.
3. **Refrigeration and temperature control**: Anything involving enzymes, live cells or long spins needs refrigeration, and a refrigerated unit costs considerably more to buy and to run. A lab that specifies ambient to save money and then needs cold spins ends up buying twice.
4. **Containment where the sample is hazardous**: Sealed buckets or a sealed rotor lid keep an aerosol inside the rotor if a tube fails, and they are what allows the rotor to be opened in a cabinet rather than at the bench. For infectious or human material this is a containment requirement, not an accessory.
5. **Rotor life and service**: Rotors are consumables on a long clock: they log cycles, they corrode, and manufacturers publish derating and retirement rules. Ask what the rotor log looks like, what the retirement policy is, and what a replacement costs, because that answer is frequently a large fraction of the machine price.

**What is the difference between rpm and relative centrifugal force?**

Rpm measures how fast the rotor turns; relative centrifugal force measures the force applied to the sample, and depends on both speed and rotor radius. Protocols specify force, so convert using the rotor's radius rather than matching rpm between machines.

**Do I need a refrigerated centrifuge?**

If you spin live cells, enzymes, nucleic acids or anything for an extended period, yes. Frictional heating during a long spin raises sample temperature noticeably even in a cool room.

**When do I need sealed buckets?**

Whenever a tube failure could release an infectious or toxic aerosol, including most work with human samples. Sealed carriers let the rotor be loaded and opened inside a biosafety cabinet.

**How long does a rotor last?**

Manufacturers set a life in cycles or years and publish derating for older rotors, and corrosion or a dropped rotor can retire one immediately. Keep a log, because the retirement rule depends on it.

**How many centrifuge types does a laboratory need?**

Usually a small instrument for tubes at low force, a general bench instrument that takes plates and larger vessels, and, where samples are temperature sensitive, refrigeration on one of them. Everything beyond that follows from a specific protocol rather than from the category list.

**What do I check before ordering a floor-standing centrifuge?**

Floor loading, clearance around the instrument, the power supply at the intended location, and whether the room can lose the heat. All four are cheaper to answer on a drawing than after delivery.

## Test tubes

Test tubes are ordered on price and regretted on fit. The dimension that matters is rarely the volume printed on the box but the outside diameter and the length, because those decide whether the tube fits your racks, your heating block, your rotor and your reader. This page sets out the material and closure decisions, and the compatibility checks that stop a delivery becoming storeroom clutter.

1. **Outside diameter and length before volume**: Racks, blocks, rotor adapters and readers are built around external dimensions. Check the diameter and length against everything the tube must sit in before ordering, because two tubes of the same nominal volume from different suppliers frequently differ enough to matter.
2. **Glass or plastic**: Borosilicate takes direct heat, resists solvents and can be flame sterilised; polypropylene survives centrifugation and freezing and does not shatter; polystyrene is clear and cheap and is limited to ambient aqueous work. Each choice rules something out, so pick from the protocol.
3. **Round, conical or flat bottom**: Conical bottoms concentrate a small pellet and make it visible; round bottoms mix better and suit culture; flat bottoms stand unaided and suit reading. The shape is a protocol decision that also determines whether the tube needs a rack to stay upright.
4. **Closure type**: Screw caps seal reliably and are slow; push caps are fast and can pop under pressure or in a freezer; septum caps allow sampling without opening. For anything frozen, shipped or heated, the closure is the part that fails, so specify it deliberately.

**What is a test tube made of?**

Usually borosilicate glass, polypropylene or polystyrene. Borosilicate takes heat and solvents, polypropylene takes centrifugation and freezing, and polystyrene is a clear low-cost option for ambient aqueous work.

**Can test tubes be autoclaved?**

Borosilicate and most polypropylene can, within the manufacturer's stated limits. Polystyrene cannot. Closures often have a lower limit than the tube, so check both.

**What is the difference between a test tube and a centrifuge tube?**

A centrifuge tube is rated for a maximum relative centrifugal force in a specified adapter and is usually conical with a sealing closure. A general test tube carries no such rating and should not be spun hard.

**Why do my tubes not fit the rack?**

Nominal volume does not fix external dimensions. Racks, blocks and rotor adapters are built for a diameter and length, and suppliers differ. Check both dimensions against every holder the tube must enter.

**Do I need certified nuclease-free tubes?**

For RNA work and sensitive amplification, yes, and the certification is separate from sterility. For general chemistry and culture it is an unnecessary cost.

**Why do my heated tubes give variable results?**

Usually a loose fit in the block. A block is machined for one tube diameter and poor thermal contact gives uneven heating. Match the block to the tube and keep one tube format per protocol.

## Beaker tongs

Beaker tongs are the cheapest item on this site and the one most often bought in the wrong size, which matters because a tong that does not grip a hot beaker properly is a burn and a spill rather than an inconvenience. The choice comes down to jaw span, coating and grip geometry, and to whether tongs are the right tool for the job at all. This page covers all three and the handling practice that should sit behind them.

1. **Jaw span against your actual glassware**: Tongs grip a range of diameters and lose their hold at both ends of it. Measure the beakers in routine use and buy for that range, with a second pair if the lab spans small and large vessels, because one pair covering everything grips nothing well.
2. **Coated or bare**: Vinyl or silicone coated jaws grip better and protect glass from scratching, and they have a temperature ceiling above which they degrade. Bare stainless or nickel plated steel handles higher temperatures and transmits heat to the hand faster. Match the coating to the hottest thing you will lift.
3. **Grip geometry and one-handed use**: Scissor-action tongs need a deliberate squeeze and hold under load; spring-action tongs close by default and tire the hand less. Whichever you buy, the beaker should be lifted with the tongs square to the vessel and close to the rim, never at an angle.
4. **Know when tongs are the wrong tool**: A full large beaker of hot liquid is not a tong job. Use a heat-resistant glove with two hands, a carrier or a trolley. Tongs are for lifting and moving modest vessels, and treating them as a universal hot-glass handler is where accidents come from.

**What are beaker tongs used for?**

Lifting and moving beakers that are hot, cold or otherwise not safe to touch, within the size range the jaws grip properly. They are not intended for large full vessels or for round-bottom flasks.

**What size beaker tongs do I need?**

Match the jaw span to the beaker diameters you actually use. Most labs need two pairs to cover a small and a large range, because a single pair sized for the middle grips neither end reliably.

**Are coated tongs better?**

They grip better and protect the glass, up to the coating's temperature limit. Above that limit, bare steel is the correct choice and a heat-resistant glove becomes more important.

**Can beaker tongs hold a flask?**

Not safely. A conical or round-bottom flask has no straight wall for the jaws to grip. Use flask tongs or a clamp designed for the shape.

## Beakers

A beaker is a container with a spout, and the differences between a good one and a poor one show up in use rather than in the catalog: whether it pours without running down the outside, whether the rim survives handling, and whether the glass takes a hotplate without cracking. This page covers the grades and forms, and what the printed marks on the side are and are not for.

1. **Borosilicate for heat, plastic for safety**: Borosilicate glass takes direct heating on a plate and repeated autoclaving. Polypropylene and polymethylpentene do not shatter and suit cold rooms, field work and anything carried; each has solvent limits and a temperature ceiling that borosilicate does not.
2. **Low form or tall form**: The low or squat form is the general-purpose beaker: stable, easy to stir, quick to evaporate from. The tall form reduces evaporation and suits titration and precipitation work. Buying only one form is why labs improvise with the wrong vessel.
3. **Spout and rim**: A well-formed spout pours cleanly and a reinforced rim resists chipping, and these are the two features that most distinguish laboratory glass from cheap glass. A chipped rim is a cut and a contamination route and should retire the beaker.
4. **Graduations are for guidance**: The molded scale on a beaker is approximate by design and is there so you know roughly where you are. Any volume that appears in a result is measured in a cylinder or a volumetric flask. Treating beaker marks as measurements is the most common volumetric error in a teaching or production lab alike.

**Can you measure volume in a beaker?**

Only approximately. The molded graduations are indicative and carry no tolerance grade. Use a graduated cylinder or a volumetric flask for any volume that forms part of a result.

**What is a science beaker made of?**

Usually borosilicate glass, which tolerates heating and solvents, or a polyolefin plastic such as polypropylene where breakage matters more than heat resistance.

**What is the difference between low form and tall form beakers?**

Low form is wider and shallower, stable and easy to stir and evaporate from. Tall form is narrower, reduces evaporation and suits titration and precipitation. Most benches need both.

**Can a beaker go on a hotplate?**

A borosilicate beaker can, ideally with a diffuser and avoiding sudden temperature changes. Plastic beakers cannot go on a hotplate at all.

**Can I measure a volume in an open beaker?**

Only approximately. Its graduations are indicative and its wide surface evaporates quickly. Where the volume has to be right, use a graduated or volumetric vessel with a stated tolerance class.

**When is a quartz beaker worth the cost?**

For temperatures beyond what borosilicate tolerates and for anything involving ultraviolet light, which ordinary glass absorbs. For everyday heating and mixing it buys nothing.

## PBMC supply

Peripheral blood mononuclear cells are bought from a supplier rather than made, and what you are really buying is a documentation chain: consented collection, a characterised donor, a validated isolation and a cold chain that did not break. The cells either work in your assay or they do not, and the difference is usually upstream of anything you can see in the vial. This page covers what to specify and what evidence to require before a purchase order goes out.

1. **Fresh or cryopreserved**: Fresh cells ship within a short window and give the most physiological response, at the cost of scheduling your experiment around a collection. Cryopreserved vials are available on demand and from a characterised lot you can return to, at the cost of viability and of a recovery step that has to be validated for your assay.
2. **Donor characterisation**: Decide which donor attributes your question actually depends on: serology status, HLA type, smoking, medication, disease state, age band and sex are all available at a price. Specifying attributes you do not need narrows the donor pool and raises the cost without improving the experiment.
3. **The consent chain**: Ask for evidence that collection was under an approved protocol with informed consent covering the research use you intend, including commercial use if that is where the work goes. A supplier that cannot evidence this is a supplier whose material may not be usable in a publication or a filing.
4. **Lot reservation and repeatability**: If the study runs over months, reserve the lot. A comparison across donors is a different experiment from a comparison across time on one donor, and discovering mid-study that the lot is exhausted forces the first when you designed the second.
5. **Cold chain and acceptance**: Require a temperature record with the shipment and define acceptance criteria before it arrives: viability on thaw, recovery, and the assay-relevant functional check. Agree in writing what happens when a shipment fails them, because that conversation is much harder afterwards.

**What are PBMC cells?**

Peripheral blood mononuclear cells: the lymphocyte and monocyte fraction isolated from blood by density gradient. They are the standard starting material for immunology, immuno-oncology and vaccine response work.

**Should I buy PBMCs fresh or cryopreserved?**

Fresh gives the most physiological response and requires scheduling around a collection. Cryopreserved is available on demand from a repeatable lot and needs a validated thaw and rest step. Most programs use cryopreserved and bridge to fresh where function is in question.

**What documentation should I ask for before I buy human PBMC material?**

Evidence of consented collection under an approved protocol covering your intended use, donor characterisation to the attributes you specified, infectious disease testing results, the isolation and cryopreservation method, and a lot certificate with viability and count.

**Can I use purchased human PBMCs for commercial research?**

Only if the donor consent and the supplier's terms cover commercial use. This is a common and expensive thing to discover late, so confirm it in writing before the first purchase rather than at publication or filing.

## Recombinant proteins

A recombinant protein is sold either from a catalog or made to order, and the two are priced and specified so differently that buyers routinely compare quotes that are not comparable. What decides both the price and whether the protein works is the expression system, the tag, the purification route and what the supplier will actually certify about activity. This page sets out those decisions and how to write a specification that gets you comparable quotes.

1. **The expression system decides the protein**: Bacterial expression is cheap, fast and gives no mammalian glycosylation or complex folding. Insect and mammalian systems cost considerably more and produce material that behaves like the native protein. A quote is only meaningful once the system is fixed, because it is the largest single price driver and the largest single quality driver.
2. **Tags are a trade, not a free addition**: An affinity tag makes purification straightforward and can change folding, activity and immunogenicity. Decide whether the tag is cleaved, what the residual scar is, and whether you need a tag-free preparation, because tag removal adds a step and a yield loss that must be in the quote.
3. **Purity is a method, not a number**: A purity claim means nothing until you know how it was measured. Ask which method produced the figure and ask for the trace. Two suppliers claiming the same purity by different methods are not making the same claim.
4. **Activity is the specification that matters**: For an enzyme or a ligand, ask what activity assay is run, on what schedule, and whether the result is lot-specific or a generic statement. A protein certified only for purity can be pure and inactive, which is the most common disappointment in this category.
5. **Scale and repeat supply**: Custom production is quoted per batch and the unit price falls sharply with scale. If the program will need more, negotiate the repeat batch and the cell bank retention at the same time as the first order, when you have the most leverage.

**What is a recombinant protein?**

A protein produced by expressing its gene in a host organism rather than purifying it from its natural source. The host, the construct and the purification together determine whether it behaves like the native protein.

**Which expression system should I choose?**

Bacterial for simple, non-glycosylated proteins where cost and speed matter; insect or mammalian where folding, post-translational modification or native activity matter. The choice dominates both price and behavior, so fix it before comparing quotes.

**Do I need a tag-free protein?**

For structural work, for immunisation and for anything where the tag could interfere with binding or activity, usually yes. For routine pull-downs and detection the tag is often useful. Removing it costs a step and some yield.

**Why does endotoxin matter?**

Bacterially expressed protein carries endotoxin that activates immune cells at low levels, so it confounds any cell-based or in vivo experiment. Specify a limit and require a lot-specific result.

**Does the expression host matter?**

Considerably. Bacterial protein carries no glycosylation and arrives with endotoxin; yeast glycosylates differently from human cells; insect and mammalian hosts are closest to native and cost more. Where activity depends on modification, the host decides whether the product works.

**What do scFvs, the single-chain format, give up?**

The constant region, and with it effector function and a long circulating life, plus more tendency to aggregate. It gains small size and easy bacterial expression, which suits imaging, intracellular binders and building blocks.

## Protein expression services

A protein expression project is a research project with a delivery date, and the single decision that determines whether it goes well is how it is staged. A fixed price for a difficult protein transfers risk to the provider, who prices that risk in; a staged contract with a genuine stop point after feasibility usually costs less and tells you sooner whether the protein is going to behave. This page covers the staging, the screening and the terms worth arguing about.

1. **Feasibility before production**: A short feasibility stage tests expression and solubility in a small number of constructs and hosts and ends with a decision rather than a delivery. It is the cheapest information you will buy on the project, and a provider unwilling to sell one is a provider proposing you fund their risk.
2. **Construct and host screening**: Screening varies the boundaries, the tag, the fusion partner and the host in parallel. The value is in breadth rather than depth, so ask how many variants the screen actually covers and how a winner is chosen, because a screen of two is not a screen.
3. **Scale-up is a separate decision**: What works in a small culture does not always survive scale-up, and a provider should say so. Price scale-up as its own stage with its own acceptance criteria rather than assuming the feasibility yield holds.
4. **Cell bank and material ownership**: Settle who owns the construct, the cell bank and the process, and where they are stored. A program that will re-order should hold or be able to obtain the bank, or the second batch is a negotiation from a weak position.
5. **Acceptance criteria in the contract**: Write the quantity, the purity method and threshold, the activity assay and the endotoxin limit into the agreement. Acceptance disputes in this category are almost always about criteria that were discussed and never written down.

**What does a protein expression service include?**

Typically construct design and synthesis, transformation or transfection, small-scale expression and solubility screening, scale-up, purification and the agreed analytics, delivered against a written specification.

**How long does a protein expression project take?**

It depends on the host and whether the protein expresses at all. Bacterial projects on a well-behaved target are short; mammalian expression, difficult targets and anything needing a stable line take substantially longer. Stage the contract so you learn which you have early.

**Should I pay a fixed price or by stage?**

By stage for anything uncertain. A fixed price for a difficult protein means the provider has priced the risk of failure into your quote whether or not it happens.

**Who owns the cell bank?**

Whatever the contract says, so read it. If you expect to re-order, ensure you own or can obtain the bank and the process description, otherwise the second batch is a sole-source negotiation.

## Protein purification

Purification is a sequence of trades between yield and purity, and every additional step costs material. A route designed around the analytics you actually need is shorter, cheaper and more reproducible than one designed to make a gel look impressive. This page covers how a route is put together, what to specify when you outsource it, and the analytics that should accompany the delivered material.

1. **Capture first, and quickly**: The first step should take the target out of the crude mixture fast, usually by affinity where a tag or a natural ligand allows it. Speed matters because proteases and oxidation are working the whole time, and a slow capture costs activity you cannot recover later.
2. **Intermediate purification removes the near neighbors**: Ion exchange or hydrophobic interaction separates the target from contaminants that survived capture, and this is where most of the real purity is won. It is also where most of the yield is lost, so the conditions deserve real development rather than a default method.
3. **Polish only if the application needs it**: Size exclusion removes aggregate and exchanges buffer and is the usual final step. It is also dilute and slow. If the application tolerates the aggregate level after the intermediate step, the polish is a step you are paying for twice, once in cost and once in yield.
4. **Endotoxin and nucleic acid removal**: Material destined for cells or animals needs endotoxin and host nucleic acid addressed explicitly, usually with a dedicated step and a lot-specific test. Treating these as incidental is the most common reason purified protein fails downstream.
5. **Buffer and format on delivery**: Specify the final buffer, the concentration, the volume and whether the material is frozen, lyophilised or shipped cold. A protein delivered in the provider's default buffer frequently needs an exchange you then perform badly.

**What is the usual order of purification steps?**

Capture to get the target out of the crude material quickly, intermediate purification to remove close contaminants, then a polish to remove aggregate and exchange buffer. Not every protein needs all three.

**How much protein will I lose?**

Every step costs yield, and the loss depends on the target and the conditions rather than on a general rule. This is the argument for the shortest route that meets the specification the application actually needs.

**Should I purify in house or outsource?**

Outsource one-offs and work that needs equipment or scale you do not have. Keep in house anything you will repeat, because the method knowledge is worth more than the first batch.

**What analytics should accompany purified protein?**

Purity with the method and trace, concentration with the method, aggregate where relevant, identity confirmation, and an endotoxin result if the material goes near cells or animals.

**Kit, column or magnetic beads for protein purification?**

A kit for screening constructs and one-off preparations of a few milligrams; a column on a system where the purification repeats, because the cost per milligram and the control of the separation are both far better.

**Can an endotoxin removal resin remove all the endotoxin?**

It reduces it rather than eliminating it, and it can bind the product too. Measure by the compendial method with an interference screen and set a limit in advance as a release criterion.

## Antibody production

Antibody production is sold as a package and delivered as a sequence of decisions, and the decisions that determine whether you get a useful reagent are made in the first few weeks: the immunogen, the host and the screen. Providers differ far more in how they screen than in how they immunise, and screening is where a project either finds the antibody you need or produces a binder that fails in your assay. This page covers the stages, what to specify, and what to own at the end.

1. **Immunogen design decides the outcome**: Whether you immunise with a peptide, a recombinant domain or the whole protein determines what the resulting antibodies recognize, and a peptide-raised antibody frequently fails against the folded protein. Choose the immunogen from the application rather than from convenience or cost.
2. **Host species and format**: The host affects affinity, the epitope repertoire and downstream compatibility with your detection reagents and your species of study. Rabbit, mouse, goat and the camelid single-domain formats each suit different applications, and the choice constrains what you can multiplex with later.
3. **Screening is the whole project**: Ask what the screen actually is and how closely it resembles your application. A binder selected on a plate against the immunogen may not work in the assay you need, so pay for a screen in the relevant format even where it costs more, because it is much cheaper than a failed reagent.
4. **Polyclonal or monoclonal**: Polyclonal serum is fast and cheap and is finite: when the animal is gone, the reagent is gone. A monoclonal or a recombinant monoclonal is renewable and consistent and costs more up front. For anything that must be reproducible for years, the renewable route is the cheaper one.
5. **Ownership and renewability**: Settle who owns the hybridoma, the sequence and the cell line, and whether you can have the variable region sequence. A sequenced antibody can be made again anywhere; a hybridoma in somebody else's freezer is a dependency with a lifetime.

**Custom polyclonal antibody or monoclonal: which should I commission?**

Polyclonal is faster and cheaper and is a finite resource tied to one animal. Monoclonal, and especially recombinant monoclonal, is renewable and consistent and costs more up front. Long-running work is cheaper on the renewable route.

**How long does antibody production take?**

It is driven by the immunisation schedule and the depth of screening rather than by the provider. Polyclonal projects are substantially shorter than monoclonal ones, and recombinant routes add sequencing and expression time at the end.

**What should I own at the end?**

Ideally the sequence of the variable regions, which makes the antibody reproducible anywhere. Failing that, the hybridoma or cell line and the right to have it made. Serum alone is a reagent with an expiry.

**Will the provider validate the antibody for my assay?**

Only if you pay for a screen in that format. The standard screen tests binding to the immunogen, which does not predict performance in tissue, in a fixed sample or in a complex lysate.

**What does an antibody discovery service include that production does not?**

Discovery finds the binder: immunisation or library selection, screening against the target, and enough characterisation to pick candidates. Production makes material from a binder you already have. A quotation that covers both should say where one ends, because the deliverable of discovery is a sequence and a small amount of protein, not a supply.

## Antibody generation

Generation is the discovery half of the antibody problem: finding binders at all, rather than producing more of one you already have. The platform decides what you can find, how fast, and what you own afterwards, and the three main routes differ far more than the marketing suggests. This page sets out what each platform is good at, where each fails, and how to stage a contract so a failure is discovered early and cheaply.

1. **Immunisation and hybridoma**: The classical route gives affinity-matured antibodies against native epitopes and works well for immunogenic targets. It is slower, depends on an animal responding, and yields a hybridoma whose sequence you should insist on having so the reagent survives the freezer.
2. **Phage and yeast display**: Display screens very large libraries quickly and in vitro, so it handles toxic and non-immunogenic targets and gives fine control over selection conditions. Binders come out as sequences, which is convenient, but affinity maturation is a separate engineered step rather than a gift from the immune system.
3. **Single B cell screening**: Sequencing paired chains from individual B cells of an immunised animal combines native pairing with direct sequence output. It is the most expensive route and the one that most reliably gives renewable, sequenced leads from a real immune response.
4. **Selection pressure is where you steer**: Whatever the platform, what you find is what you selected for. Counter-selection against close homologues, selection at the pH or in the buffer the application uses, and screening in the assay format all change the leads. Specify these rather than leaving them to the provider's default.
5. **Stage the contract at the first lead**: Break the project at the point where leads exist and are characterised, before any engineering or production. That is the natural decision point, and a contract without it commits you to spending on a campaign that may have found nothing useful.

**What is the difference between antibody generation and antibody production?**

Generation is discovering binders against a target; production is making more of an antibody that already exists. Generation projects are discovery work with real failure risk, which is why they should be staged.

**Which of the antibody discovery platforms should I choose?**

Hybridoma for immunogenic targets where affinity matters and budget is limited; display for toxic, conserved or non-immunogenic targets and where selection control matters; single B cell where you want native pairing and sequenced leads and can afford it.

**Can a generation campaign fail?**

Yes, and it is the reason to stage the contract at the first characterised lead. Difficult targets, highly conserved epitopes and poor immunogens all produce campaigns with no usable binder.

**Do I get the antibody sequence?**

Only if the contract says so. Ask for the variable region sequences of every lead you pay to characterise, because that is what makes the reagent renewable independently of the provider.

**When should I commission a custom antibody?**

When no catalog reagent has been validated for your target and application and you have already spent weeks testing candidates. Write the acceptance criteria into the contract: the application, the species and the control that has to pass.

## Cell line development

Cell line development produces the thing every later stage depends on, which is why the evidence generated along the way matters as much as the productivity number at the end. A line that expresses well but cannot be shown to be clonally derived, or that loses expression over the generations a campaign needs, is a line you will develop twice. This page covers the stages, the evidence that has to accompany each, and the questions that separate providers.

1. **Host and vector selection**: The host line and the expression vector constrain productivity, product quality and what regulatory history you inherit. Providers offer their own hosts under license, and the license terms travel with your program, so read them before the science starts.
2. **Pool generation and early assessment**: Transfected pools give an early read on whether the molecule expresses and what its quality attributes look like, long before clones exist. Material from a pool is often enough for early studies, and generating it early removes months from a program's critical path.
3. **Single cell cloning and clonality evidence**: The line must be derived from a single cell, and that has to be evidenced rather than asserted, normally by imaging at the point of deposition with a retained record. Regulators ask for this evidence directly, so confirm what the provider captures and what you receive.
4. **Stability over the generations you need**: Expression and product quality are assessed over a generation number that covers your intended manufacturing scale plus a margin. A stability study that stops short of what the process needs is a study you will repeat.
5. **Cell banking**: A research or master bank is laid down with identity, sterility, mycoplasma and adventitious agent testing to a defined panel. Establish who holds the bank, where the backup is stored, and what your rights to it are if the relationship ends.

**What does cell line development deliver?**

A clonally derived, documented cell line expressing your product, with clonality evidence, a stability assessment over the required generations, a cell bank with its testing package, and the process description.

**Why does clonality evidence matter?**

Regulators expect assurance that the production line descends from a single cell, and they ask for the evidence rather than the assertion. Imaging at deposition with a retained record is the usual form, and it is far easier to capture at the time than to reconstruct.

**Can I use pool material before stable cell lines are ready?**

Usually yes, and doing so pulls months out of a program. Pool material gives an early read on expression and quality while cloning proceeds in parallel.

**Who owns the cell line?**

The contract decides, and the host and expression system licenses sit underneath it. Confirm both before starting, because license obligations attach to the program and can outlast the provider relationship.

## Bioreactors

A bioreactor is bought for a scale and used for a process, and the mismatch between the two is where most disappointment sits. The vessel volume in the catalog is not the working volume, the control system decides what the process can actually do, and the consumables decide what it costs to run. This page separates the vessel decision from the control decision and covers what to establish before an order.

1. **Working volume, and the turndown ratio**: The usable range of a vessel is narrower than its nominal volume at both ends, because probes must stay submerged and headspace must be preserved. Ask for the minimum and maximum working volume, since a vessel that cannot run small cannot be used for development.
2. **Single use or stainless**: Single use removes cleaning and sterilisation validation and shifts cost into consumables and into a supply chain you do not control. Stainless costs more up front, more in utilities and validation, and is cheaper per run at high utilization. Utilization is what decides this, not preference.
3. **Mixing and gas transfer**: Impeller type and sparging arrangement set the oxygen transfer and the shear the culture experiences, and mammalian and microbial processes want very different answers. Specify from the organism and the process, because retrofitting a different impeller is rarely straightforward.
4. **Sensors and control**: Decide which parameters are controlled rather than merely monitored, and which measurements are in line rather than sampled. Single use vessels constrain sensor choice more than stainless does, and that constraint is easier to accept before the purchase than after.
5. **Data and integration**: Establish how the batch record leaves the controller and whether it can reach your system of record without manual transcription. A process whose data lives only in the controller is a process whose history is hard to defend later.

**What size bioreactor do I need?**

Size from the working volume your process needs, not the nominal vessel volume, and confirm the minimum working volume as well as the maximum. A vessel that cannot run at small volumes is not usable for development.

**Single-use bioreactors or stainless steel?**

Single use where utilization is moderate, changeover is frequent or cleaning validation is a burden. Stainless where utilization is high and the process is stable. The deciding factor is how often the vessel runs, not the purchase price.

**Bioreactor, fermenter: what is the difference?**

The terms overlap. In common usage a fermenter runs microbial cultures at high oxygen demand and high shear tolerance, while a bioreactor often means a mammalian cell system with gentler mixing. The mixing and sparging design is what actually differs.

**Can I scale a process between different manufacturers' vessels?**

Yes, but it is harder. Geometric differences change mixing and gas transfer, so more of the process has to be re-characterised. Staying within one geometry family simplifies scale-up considerably.

**Can one vessel serve microbial and mammalian culture?**

Rarely. Microbial cultures need high agitation, gas transfer and cooling; animal cells are shear-sensitive and slow. The organism decides the vessel before the volume does.

**Bioreactor cost: what actually costs money, and what moves the bioreactor price?**

The control system, the probes and the software rather than the vessel. Ask what is controlled against merely monitored, how probes are calibrated and sterilised, and whether data export in a usable form.

**How is pressure measured in a disposable flow path?**

Either with an in-line disposable transducer irradiated with the assembly and calibrated at manufacture, or with a clamp-on sensor reading through the tube wall, which avoids fluid contact and is less accurate as the tubing creeps.

**What range should a pressure transducer sensor have?**

The range the step actually reaches. Accuracy is a fraction of full scale, so an oversized transducer resolves a low pressure poorly.

## Custom antibodies

Most custom antibody projects are started because a catalog reagent failed, and a good number of them fail the same way for the same reason: the specification described the target rather than the application. An antibody is only ever fit for a purpose, and the purpose has to be written down before the immunogen is chosen. This page covers when a custom order is genuinely the right answer, and what the specification needs to contain.

1. **Rule the catalog out properly first**: Before commissioning, establish whether existing reagents genuinely fail or were used outside their validated application. A catalog antibody validated for one application and used in another is the most common false negative, and finding that out costs a great deal less than a custom project.
2. **Write the application into the specification**: State the assay, the sample type, the fixation and the species. An antibody that must work on fixed tissue, on a denatured blot and in a live-cell assay is three different reagents, and a provider that agrees to all three without comment is not reading the request.
3. **Choose the epitope deliberately**: Where you raise against determines what you can distinguish: an isoform, a modification state, a species orthologue or a domain. Say what must be discriminated and what must be recognized across, and require a counter-screen that tests it.
4. **Plan validation before the project starts**: Decide what evidence will convince you the antibody works, including a genuine negative control such as a knockout or knockdown sample. Projects without a defined acceptance test end in an argument about whether a band is the right band.
5. **Secure renewability**: Ask for the sequence or the clone. A custom polyclonal is a finite reagent, and a program that will run for years should pay the extra for something that can be made again.

**When should I order a custom antibody?**

When no catalog reagent exists for the target, when you need to distinguish an isoform, modification or species that existing reagents cannot, or when you have genuinely established that available reagents fail in your validated application.

**How long does a custom antibody take?**

It is set by the immunisation and screening schedule rather than the provider. Polyclonal projects are considerably shorter than monoclonal or recombinant ones, which add cloning, sequencing and expression.

**What should the specification contain?**

The target and the epitope region, what must be discriminated and what must be recognized across, the assay and sample type, the host and format, the screening format, and the acceptance criteria for validation.

**How do I validate a custom antibody?**

In the application it was commissioned for, with a real negative control such as a knockout, knockdown or non-expressing sample. A signal at the right size is not validation on its own.

**How do I judge research antibodies before buying them?**

Look for a genetic control, validation in your application and species, a disclosed immunogen and a persistent identifier you can search the literature with. The number of products against a target measures popularity, not evidence.

## Column chromatography

A chromatography column is sold on its dimensions and its packing and bought for a separation, and the supplier cannot know whether it will achieve yours. What determines the result is the match between the stationary phase chemistry and the difference you are trying to exploit, with particle size, pore size and bed dimensions deciding how much resolution you buy for how much pressure and time. This page sets out those trades so a column order follows the method rather than the catalog.

1. **Phase chemistry follows the difference you exploit**: Reversed phase separates on hydrophobicity, ion exchange on charge, size exclusion on hydrodynamic size, and affinity on a specific interaction. Choose the mode from the property that actually differs between your target and the contaminants, because no dimension compensates for the wrong mode.
2. **Particle size trades resolution against pressure**: Smaller particles give sharper peaks and demand higher pressure and a system built for it. Putting a fine-particle column on a system that cannot deliver the pressure produces a slow separation and a damaged column, so match the column to the instrument as well as to the method.
3. **Pore size must admit the molecule**: For proteins and other large molecules the pore has to be large enough for the analyte to enter the particle, or the surface area is unavailable and capacity collapses. This is the single most common cause of a protein method behaving nothing like the small molecule method it was adapted from.
4. **Dimensions set capacity and time**: Bed length drives resolution and back pressure; diameter drives loading capacity and solvent consumption. Scale these deliberately from whether the run is analytical or preparative, because using an analytical column preparatively overloads it and destroys the separation.
5. **Lot to lot reproducibility**: Ask whether the supplier publishes a test chromatogram per lot and what its acceptance criteria are. A method that has to survive a column change needs a supplier whose packing is controlled, and that information is available if you ask for it.

**How do I choose a chromatography column, including ion chromatography columns?**

Start from the property that differs between your target and the contaminants, which fixes the separation mode. Then set pore size to admit the analyte, particle size from the resolution and pressure your system supports, and dimensions from whether the run is analytical or preparative.

**What does particle size change?**

Smaller particles give sharper peaks and higher back pressure. The column must be matched to a system that can deliver that pressure, otherwise the method runs slowly and the column suffers.

**Why does pore size matter for proteins?**

A protein too large to enter the pores cannot reach the internal surface, so the effective capacity and retention collapse. Wide-pore packings exist for exactly this reason.

**How long does a column last?**

It depends on sample cleanliness, mobile phase aggressiveness and whether a guard column is used. Track injections and back pressure rather than time, since pressure trend is the signal that a column is failing.

**Should I pack my own or buy a prepacked column?**

Only where the resin, bed height or volume is unusual and the skill exists to pack evenly and test the bed. Otherwise a prepacked column arrives tested with an efficiency specification, which is worth the premium for consistency between runs.

## Gas chromatography

Gas chromatography separates what can be vaporised without decomposing, and almost every decision in specifying a system follows from the detector, because the detector decides what you can see and at what level. The rest of the instrument then has to support it: an inlet that introduces the sample without discrimination, a column that resolves the analytes, and a carrier gas supply that does not run out or introduce contamination. This page covers the choices and what to ask an outsourced provider.

1. **The detector decides the instrument**: A flame ionisation detector is robust and near-universal for organics; an electron capture detector is exquisitely sensitive to halogenated compounds; a mass spectrometer gives identification as well as quantitation and costs and demands far more. Fix the detector from the analytes and the required detection level first.
2. **Inlet and sample introduction**: Split, splitless, on column and headspace introduce very different amounts of sample and treat thermally labile analytes very differently. Discrimination at the inlet is a common and invisible source of quantitative error, so choose the mode from the analyte rather than the convenience.
3. **Column phase and dimensions**: Stationary phase polarity, film thickness, internal diameter and length set resolution, capacity and run time. A longer, thinner column resolves better and runs slower; a thicker film retains volatiles better and bleeds more at high temperature.
4. **Carrier gas supply**: Helium supply and price have been volatile, and hydrogen and nitrogen are both viable alternatives with consequences for efficiency and for safety. If hydrogen is chosen, the laboratory ventilation and detection arrangements are part of the purchase rather than an afterthought.
5. **Data system and records**: The chromatography data system holds the raw data and its audit trail, and it is where a regulated laboratory's integrity questions land. Confirm how raw data is retained, who can reprocess it and what the audit trail captures before the system is chosen.

**What can gas chromatography analyse?**

Compounds that can be vaporised without decomposing, which covers most volatile and semi-volatile organics. Thermally labile, very polar or non-volatile compounds usually need derivatisation or a liquid phase technique instead.

**Which detector should I choose?**

Flame ionisation for general organics, electron capture for halogenated compounds at low levels, and mass spectrometry where identification as well as quantitation is needed. The detector determines the detection level and most of the cost.

**Helium, hydrogen or nitrogen?**

Helium is conventional and its supply has been unreliable. Hydrogen gives faster, more efficient separations and requires ventilation and leak detection arrangements. Nitrogen is cheap and slower. The decision affects the laboratory as well as the instrument.

**What should I ask a testing laboratory?**

Which method by designation, whether it is within their accredited scope, the reporting limit in your matrix, the turnaround, and whether they will supply the raw chromatogram with the certificate.

**What does a gas chromatography machine really cost to run?**

Carrier and detector gases or generators, columns as consumables, standards, a service contract and an operator's time. The gas installation alone is frequently comparable with the cheaper modules of the instrument.

## Size exclusion chromatography

Size exclusion is the gentlest separation available and the easiest to set up badly, because its resolution depends on parameters that feel like details: how much sample you load relative to the bed, how fast you run it, and whether the fractionation range actually brackets your species. Get those right and it separates aggregate from monomer cleanly; get them wrong and it produces one broad peak. This page covers the three that matter.

1. **Fractionation range must bracket your species**: A resin separates only within its stated range. Anything above elutes in the void and anything below elutes together at the end, so aggregate, monomer and fragment all have to fall inside the range if you intend to distinguish them. Choosing the resin from the species you need resolved is the first decision.
2. **Load a small fraction of the bed volume**: Resolution falls steeply as the loaded volume rises, and this is the dominant error in practice. Size exclusion is a low-capacity technique by design, so a run that loads too much to save time produces a chromatogram that cannot be interpreted.
3. **Flow rate and bed height**: Slower flow and a taller bed both improve resolution and both cost time. For a polishing step the trade usually favors resolution; for a buffer exchange, speed is fine because no separation is being attempted.
4. **Buffer prevents unwanted interactions**: The mode is supposed to be purely size based, and it stops being so when the protein interacts with the matrix. Adequate ionic strength usually suppresses this; a protein tailing badly on a size exclusion column is usually telling you the buffer is wrong rather than the resin.
5. **Calibrate before you quote a size**: Apparent molecular weight from a size exclusion column is relative to the standards used and to the shape of the molecule. An elongated protein reads larger than a globular one of the same mass, so treat the figure as a comparison rather than a measurement unless the column is coupled to light scattering.

**What does size exclusion chromatography separate?**

Molecules by hydrodynamic size, with larger species eluting first because they are excluded from the pores. Within a resin's fractionation range it is the standard way to separate aggregate from monomer.

**How much sample can I load on a size exclusion column?**

A small fraction of the column volume, and resolution falls quickly as that fraction rises. It is a low-capacity technique, which is why it is used for polishing rather than capture.

**Is size exclusion chromatography the same as gel filtration chromatography?**

Effectively yes. Gel filtration is the older name, used most often for aqueous protein separations, while size exclusion is the general term including organic phase work.

**Can I use it to measure molecular weight?**

Only relative to the standards and the shape of the molecule, since elongated proteins read larger than globular ones of equal mass. For an absolute value, couple the column to multi-angle light scattering.

**Is a size exclusion aggregate figure definitive?**

No. Aggregates can form or dissociate on the column and the figure depends on the method, so fix the conditions and confirm with an orthogonal technique where the number is reported.

## Bioinformatics tools

Almost every analysis a laboratory needs can be done with free, well-maintained tools, so the real question is never whether the software exists but who will run it, keep it current and make the result reproducible in two years. That is a staffing and infrastructure decision dressed as a software decision, and it is where commercial platforms earn their price or fail to. This page separates the two and covers what to check before committing to either.

1. **Start from the analysis, not the platform**: Write down the analyses the lab actually performs and how often. Most labs do a small number of things repeatedly, and a platform bought for the breadth of its catalog is usually bought for features nobody will use.
2. **Count the true cost of free**: Open tools cost nothing to license and require someone competent to install, update, parameterise and interpret them. If that person exists and has the time, free tools are usually better because they are what the literature uses. If they do not, a hosted platform is cheaper than the post that would be needed.
3. **Reproducibility is a requirement, not a feature**: A result you cannot reproduce is a result you cannot defend. Require versioned tools, recorded parameters and a captured environment, whether that comes from a workflow manager and containers or from a platform that records provenance for you.
4. **Data movement and storage dominate at scale**: Sequencing output is large, and moving it repeatedly between storage and compute is often the slowest and most expensive part of the workflow. Decide where the data lives before choosing where the analysis runs, because the second follows the first.
5. **Check the exit**: Ask how results and raw data leave a hosted platform, in what formats, and what happens at the end of a contract. A platform that holds your analysis history in a form you cannot export is a platform you cannot leave.

**Do I need to pay for bioinformatics tools?**

Rarely for the analysis itself, since the widely used tools are free and open. You pay for the people, the compute and the reproducibility infrastructure, or you pay a platform to provide those instead.

**What makes a bioinformatics result reproducible?**

Recorded tool versions, recorded parameters, a captured software environment and the input data identified unambiguously. Workflow managers with containers give this, and some platforms record it automatically.

**Cloud or local compute?**

Follow the data. If the data is generated locally and stays local, local compute avoids transfer costs and delays. If it arrives from a sequencing provider in the cloud, analyzing it there is usually faster and cheaper.

**How do I avoid depending on one person?**

Require that every routine workflow is written down, version controlled and demonstrably runnable by a second person. This is a management decision rather than a software one and it is the one that actually protects the lab.

## Laboratory automation software

Automation software is two products sold as one. Underneath sits a driver layer that knows how to talk to each instrument, and above it a scheduler that decides what runs when and what happens when something fails. Buyers evaluate the scheduler because it demos well, and projects fail in the driver layer because the instrument on the bench is a model nobody has written for. This page covers both, and the error handling that decides whether a walk-away run is genuinely walk-away.

1. **Bring the instrument list to the first call**: Ask specifically which of your makes and models have supported drivers, at what firmware, and what a new driver costs and takes. An unsupported instrument is either an integration project or an exclusion from the workflow, and both are better known before the contract.
2. **Scheduling, and what it optimises**: A real scheduler manages resource conflicts, timed dependencies and variable durations, and the interesting question is what it does when a step overruns. Ask whether timings are guaranteed or best effort, because assays with time-critical incubations need the first.
3. **Error recovery is the whole value**: Ask what happens when a tip is missed, a plate is not where it should be, or an instrument reports a fault mid-run. Whether the system can pause, let a person intervene and resume without losing the batch is what separates genuine walk-away operation from supervised automation.
4. **Simulation before hardware**: The ability to develop and dry-run a method without occupying the deck saves an enormous amount of instrument time and is a sign of a mature product. Ask to see a method built and simulated during the evaluation rather than watching a prepared demonstration.
5. **The boundary with the system of record**: Decide what the automation layer holds and what it hands to the laboratory information system, and make sure sample identity crosses that boundary without transcription. A run whose results cannot be traced to registered samples has automated the pipetting and broken the record.

**What does laboratory automation software do?**

It drives instruments through a defined method and schedules the resources, timings and dependencies of a run, handling errors and recording what happened. The driver layer talks to instruments; the scheduler decides the order.

**Will it work with my existing instruments?**

Only where drivers exist for those makes, models and firmware versions. Ask for the list explicitly, and for the cost and timeline of any driver that has to be written.

**Does automation software replace a LIMS?**

No. The automation layer runs the instruments; the laboratory information system holds sample identity, methods, results and the audit trail. They meet at a boundary that must carry sample identity without transcription.

**What should I test during an evaluation?**

One of your own methods, on your own instruments, end to end, including an induced failure so you can see how recovery works. Measure the whole cycle time rather than the run time.

## ELISA kits

An ELISA kit is a validated method in a box, and the validation is what you are buying. Two kits for the same analyte can disagree substantially because they use different antibody pairs recognising different epitopes, are calibrated against different standards, and were validated in different matrices. This page covers how to read a supplier's data sheet properly and what to ask for before committing a study to a kit.

1. **Format follows the analyte**: Sandwich assays need two non-competing epitopes and give the best specificity for proteins; competitive formats handle small molecules and short peptides where two antibodies cannot bind at once. A prostaglandin and a biogenic amine are both far too small to present two epitopes, which is why kits for them are competitive whatever the supplier, and why their standard curves run downwards. The format is determined by the analyte rather than chosen for convenience.
2. **Check the sample type by name**: A kit validated in serum is not validated in plasma with a particular anticoagulant, in culture supernatant or in tissue lysate. Matrix effects are real and large, so confirm your exact sample type appears in the validation and, if it does not, plan a spike and recovery experiment.
3. **The standard decides what the number means**: Kits calibrated against recombinant standards and kits calibrated against an international reference material give different absolute values. Acute phase proteins and the peptide hormones are the analytes where a recognized reference material exists, so two suppliers' kits for them can be compared; for most cytokines no such material is in general use and the calibrator is the supplier's own recombinant protein. Where results must be compared across studies or over years, prefer a kit traceable to a recognized reference material and record which was used.
4. **Sensitivity, range and precision**: Look for the working range rather than the lowest detectable figure, and for within-run and between-run precision at concentrations spanning that range. A sensitivity figure on its own says nothing about whether the assay is reliable where your samples sit.
5. **Lot variation and bridging**: Ask whether the supplier provides lot-specific standard curves and whether lots are bridged. For a study running across lots, order enough of one lot or plan a bridging experiment, because an unbridged lot change mid-study is a confound.

**Why is a PGE2 ELISA kit or a histamine ELISA kit competitive rather than a sandwich?**

Because a sandwich assay captures the analyte between two antibodies binding different epitopes, and a molecule of a few hundred daltons cannot present two at once. A competitive assay instead measures displacement of a labeled analogue, so its standard curve runs downwards and its working range is narrower. The consequence for these analytes is practical: they are more sensitive to matrix than a sandwich assay and often need an extraction or a fixed dilution, so the sample preparation in the protocol is not optional.

**Why do an HMGB1 ELISA kit and an IL6 ELISA kit give different values between suppliers?**

Different antibody pairs recognize different epitopes and different forms of the analyte, and each kit is calibrated against its own recombinant standard rather than a shared reference material. For nuclear proteins released on damage and for the cytokines this is the normal situation rather than a fault, so absolute values are kit specific and the kit and lot should be recorded with the result.

**What does research use only mean?**

The kit is labeled for laboratory research and has not been reviewed for diagnostic use, so it may not be used to inform clinical decisions regardless of how well it performs technically.

**Should I validate a mouse IgG ELISA kit or a human IgG ELISA kit myself?**

Yes, at least spike and recovery plus dilution linearity in your own matrix, and a quality control sample on every plate. Immunoglobulin kits deserve one extra check: confirm the species specificity in your sample, because a reagent raised against one species' immunoglobulin frequently sees another's, and a culture supernatant containing serum is exactly where that shows up. Supplier validation was performed in their matrix, not yours.

## Laboratory equipment

Laboratory equipment is bought on a capital budget and lived with on a revenue one, and the gap between those two budgets is where most regret sits. Service, consumables, calibration, qualification and eventual disposal usually exceed the purchase price over an instrument's life, and none of them appear on the quotation. This page covers how to evaluate the whole cost, what a service contract should say, and when refurbished equipment is a good decision rather than a false economy.

1. **Ask for the consumables list before the quote**: Many instruments are sold close to cost and earn their margin on proprietary consumables. Ask what must be bought from the manufacturer, what the current prices are, and whether third party alternatives void support, because that answer can reverse a purchase decision.
2. **Service contract or time and materials**: A contract buys predictable cost and a response commitment; time and materials is cheaper when an instrument is reliable and not on a critical path. Decide from how much downtime actually costs you rather than from the discount offered.
3. **Response time is the specification that matters**: Ask what response time means contractually, whether it is to a telephone call or to an engineer on site, what the parts availability commitment is, and whether a loan instrument is provided. A cheap contract with a slow response is not insurance.
4. **Installation, qualification and training**: Confirm what installation includes, whether qualification documentation is provided for a regulated environment, and how much training comes with the instrument. Retrofitting qualification onto an installed instrument is far more expensive than buying it with the machine.
5. **End of support and disposal**: Ask how long the manufacturer commits to parts and software support, because an instrument whose support ends is an instrument with a retirement date. For anything with refrigerant, radioactive sources or hazardous components, establish the disposal route before purchase.

**Lab equipment services: a contract, or time and materials?**

Where downtime is expensive or the instrument is on a critical path, yes. Where the instrument is reliable, easily replaced or not urgent, time and materials is usually cheaper. Decide from the cost of downtime rather than the discount.

**Is refurbished laboratory equipment or a lab equipment parts swap worth it?**

From a specialist who tests, replaces wear parts and warrants the instrument, frequently yes. Ask what was replaced, what warranty applies, whether the manufacturer will support it and whether software licenses transfer.

**What does total cost of ownership include for microbiology laboratory equipment?**

Purchase, installation, qualification, training, consumables, service, calibration, utilities, floor space and disposal. Consumables and service typically dominate over an instrument's life.

**What happens to laboratory equipment servicing when a manufacturer ends support?**

Parts and software updates stop, and the instrument becomes progressively harder to keep qualified. Ask for the support commitment in years before purchase, especially for anything with embedded software.

## Chemistry lab equipment

A new chemistry laboratory is usually equipped in the wrong order, because the exciting purchases are instruments and the purchases that determine whether the laboratory can operate are containment, services and storage. Fitting out well means spending the first part of the budget on things nobody photographs. This page sets out an order that works and the decisions within each stage.

1. **Containment and services first**: Fume hoods, ventilation, gas and vacuum lines, emergency equipment and compliant storage for flammables and corrosives determine what chemistry the laboratory may perform at all. These are building works with long lead times and they cannot be retrofitted around a bench full of instruments.
2. **Heating, cooling and stirring**: Hotplate stirrers, heating mantles, chillers, a rotary evaporator and a vacuum source cover the great majority of synthetic operations. Buy more stirrer positions than you think you need, because they are the bottleneck in every working synthetic laboratory.
3. **Weighing and measuring**: An analytical balance on a proper bench, a precision balance for bulk, and volumetric ware at a grade matched to the work. Site the balance before the bench layout is fixed rather than finding a corner for it afterwards.
4. **Separation and purification**: Flash chromatography, a good supply of columns and solvents, and rotary evaporation with adequate condenser capacity. This is where throughput is won or lost, and under-specifying the evaporator is the classic mistake.
5. **Analysis last, and shared where possible**: Spectroscopy and chromatography instruments are the largest purchases and are often better accessed through a shared facility until the workload justifies ownership. An instrument used occasionally is a maintenance liability rather than an asset.

**What should a chemistry laboratory buy first?**

Containment and services: fume hoods, ventilation, gas and vacuum, emergency equipment and compliant storage. These have long lead times, they are building works, and they determine what chemistry is permitted at all.

**How many stirrer hotplates do we need?**

More than the initial estimate. Stirring positions are the standing bottleneck in synthetic laboratories, and they are among the cheapest items on the list.

**Should we buy our own analytical instruments?**

Only when workload justifies it. Occasional use is usually better served by a shared facility or an external laboratory, because an underused instrument still needs service, calibration and someone who knows it.

**What labelling is required on chemical containers?**

Containers must identify the chemical and communicate its hazards under the hazard communication standard, and the laboratory's chemical hygiene plan should set out how that is applied to working containers as well as to stock.

**Is a hot plate set to a temperature the same as a solution at it?**

No. Without an external probe controlling on the medium, the plate's set point is not the solution's temperature. For anything temperature-sensitive, control on the medium and record it with a calibrated probe.

**Why do my sonication results vary?**

Because time, amplitude and cooling are usually not fixed, and the energy heats and shears the sample. Establish the regime on a test sample and keep it constant; over-sonication degrades what you were releasing.

## Science lab equipment

Equipping a laboratory from nothing rewards restraint. The instruments that get used every day are unglamorous and cheap; the ones that look impressive in a start-up plan are frequently used monthly and would have been better accessed elsewhere for the first year. This page sets out a core list that almost every wet laboratory needs, the equipment usually better shared, and how to decide between the two.

1. **Start with what is used daily**: Pipettes and tips, a balance, a pH meter, a centrifuge, a vortex, water baths or heat blocks, a fridge and freezer, and adequate storage. These cost a fraction of one large instrument and they determine whether the laboratory can work at all.
2. **Then the containment your work requires**: A fume hood for chemistry, a biosafety cabinet for biological work at the relevant containment level, and the emergency equipment that goes with either. What you handle decides which, and this is a compliance question rather than a preference.
3. **Cold storage is bought twice**: Laboratories consistently underestimate freezer capacity and then buy an emergency second unit at a bad moment. Size generously, plan the power and heat load, and put an alarm and a documented failure plan on anything holding irreplaceable material.
4. **Share the large instruments at first**: Sequencers, mass spectrometers, imaging systems and flow cytometers are usually available through core facilities or external providers. Using them that way for a year tells you what you actually need before the capital is committed.
5. **Buy service and consumables into the plan**: Every instrument carries recurring cost, and a fit-out plan that lists only capital will be short within months. Put calibration, service and consumables into the operating budget from the start.

**What equipment does a new laboratory need first?**

The everyday items: pipettes and tips, a balance, a pH meter, a centrifuge, a vortex, heating and cooling, a fridge and freezer, and storage, plus whatever containment the work requires. These decide whether the laboratory can function.

**Should a new lab buy large instruments?**

Usually not in the first year. Core facilities and external providers give access to sequencers, mass spectrometers and imaging systems, and a year of that tells you what the real requirement is before capital is committed.

**How much freezer capacity do we need?**

More than the initial estimate, with alarms and a documented failure plan for anything irreplaceable. Under-sizing cold storage is the most reliably repeated mistake in laboratory fit-outs.

**Is refurbished equipment suitable for a new lab?**

For robust mechanical equipment from a specialist who warrants it, yes, and it stretches a first budget considerably. Be more cautious with instruments whose value depends on software that may lose support.

**How should I specify equipment for a new laboratory?**

From the protocols, not the catalog. List what the bench will run, then what each protocol needs, then find the items. Services and long-lead items first, consumables and small equipment last but budgeted.

**Can I use a clinical diagnostic instrument for research, or the reverse?**

The label is part of the specification. An instrument labeled for research use may not be used to report a clinical result, and a clinical instrument carries obligations. Check the labeling before buying across that boundary.

## Optical microscopes

The stand carries the name and the objectives decide the image. A modest frame with good objectives outperforms an expensive frame with ordinary ones every time, and yet quotations are compared on the frame because that is where the model number sits. This page covers where the money should go, which contrast method your samples need, and what to check before committing to one manufacturer's optical system.

1. **Objectives first**: Numerical aperture sets resolution and light gathering; the correction class sets how flat and how color-accurate the field is. Buy the best objectives you can afford at the magnifications you will actually use, and fewer of them, rather than a full turret of ordinary ones.
2. **Match the contrast method to the sample**: Brightfield suits stained sections; phase contrast and differential interference contrast reveal unstained living cells; fluorescence needs labeled specimens and a matched filter and illumination set. The method has to be decided before the microscope, because retrofitting some of them is expensive or impossible.
3. **Illumination and filters**: For fluorescence, the light source and the filter sets are as important as the objective and are where a system ages fastest. Confirm which fluorophores the supplied filters actually pass, rather than accepting a generic set that half-matches your dyes.
4. **Camera and the intended output**: A camera for documentation is a different purchase from one for quantitative imaging, where sensitivity, pixel size matched to the objective, and linearity all matter. Decide what the images are for before choosing, because a mismatched camera throws away the resolution the objective delivered.
5. **Ergonomics and who uses it**: If somebody sits at the instrument for hours, eyepiece height, stage position and focus placement are a genuine specification. A microscope that is uncomfortable is a microscope that gets used badly and briefly.

**What matters most when buying an optical microscope?**

The objectives. Numerical aperture and correction class determine resolution, flatness and color accuracy, and a modest stand with good objectives beats an expensive stand with ordinary ones.

**Upright or inverted?**

Upright for slides and sections, inverted for cells in dishes and flasks because the optics look up through the vessel base. The sample vessel decides it and no accessory converts one to the other.

**Do I need phase contrast?**

If you look at unstained living cells, yes, or differential interference contrast where relief matters. Brightfield alone shows very little in an unstained transparent sample.

**Can I mix objectives, or any microscope lens, between manufacturers?**

Sometimes mechanically, but objectives are corrected as part of an optical system and mixing often degrades the image in ways that are hard to trace. Treat the first purchase as a commitment to that system.

## BglII and restriction enzymes

BglII is a six-base cutter that appears in a great many cloning strategies, and like every restriction enzyme it is sold by units rather than by mass, which makes price comparison between suppliers meaningless until you read how each defines a unit. This page covers what to check on the data sheet, the conditions that produce unexpected cutting, and the quality specifications that separate an enzyme suitable for a difficult digest from one that is not.

1. **The unit definition is supplier specific**: A unit is the amount that digests a stated quantity of a stated substrate in a stated time under stated conditions, and suppliers vary all four. Comparing price per unit between two catalogs without comparing the definitions underneath is how buyers convince themselves one enzyme is cheaper.
2. **Buffer and the double digest**: Every enzyme has an optimum buffer, and a double digest needs conditions where both work acceptably. Suppliers publish compatibility tables, and where no shared condition exists the answer is sequential digestion with a clean-up between rather than a compromise buffer that halves both activities.
3. **Star activity and how to avoid it**: Relaxed specificity under the wrong conditions produces cuts at similar but incorrect sites, which is the usual cause of an unexplained extra band. Excess enzyme, extended incubation, high glycerol from a large enzyme volume and the wrong ionic strength are the common triggers, and all four are avoidable.
4. **Methylation sensitivity**: Site methylation can block cutting entirely, and whether it does depends on the enzyme and the source of the DNA. Check the supplier's methylation table against your plasmid's host strain before concluding that a digest failed for another reason.
5. **Quality specifications that matter**: Look for stated freedom from contaminating nuclease and from residual endonuclease, and for a ligation and recutting assay result. Those tests are what tell you an enzyme will work in a cloning workflow rather than merely cut on a gel.

**What does BglII cut?**

It recognizes a six-base palindromic site and leaves a four-base overhang compatible with several other enzymes, which is why it appears frequently in cloning strategies. Check the supplier's data sheet for the exact site and compatible overhangs.

**Why is my digest incomplete?**

Common causes are insufficient units for the DNA quantity, the wrong buffer for a double digest, site methylation in the host strain, or a degraded enzyme aliquot. Check the methylation table and run a fresh aliquot before anything else.

**What is star activity?**

Cutting at sites similar to but not identical with the recognition sequence, caused by excess enzyme, long incubation, high glycerol or the wrong ionic strength. It presents as unexpected extra bands.

**How do I compare enzyme prices between suppliers?**

Read each supplier's unit definition first, because the substrate, quantity, time and conditions differ. Price per unit is not comparable until the definitions are.

## Proteinase K

Proteinase K is used to destroy protein, which includes the nucleases that would otherwise destroy your nucleic acid, and it works in conditions that inactivate most other enzymes. That robustness is exactly why it has to be inactivated or removed deliberately, because an enzyme that survives detergent and heat will happily digest the polymerase in the next step. This page covers grade selection, the activity specification, and getting rid of it afterwards.

1. **Grade is about what it does not contain**: Molecular biology grade is certified free of detectable nuclease activity, which is the whole point when the enzyme is being used to protect nucleic acid. General or technical grade is cheaper and is appropriate for protein removal where downstream nucleic acid integrity does not matter.
2. **Activity units, not mass**: Preparations are specified in activity units per milligram against a defined substrate, and lot-to-lot specific activity varies. Dose from the units rather than the mass, particularly when switching supplier or lot, or a protocol calibrated on one preparation will under or over digest on another.
3. **Conditions that raise activity**: Activity increases with temperature within the working range and is enhanced by denaturants and detergents, which is why lysis buffers pair it with them. Calcium stabilises the enzyme, so chelators in the buffer reduce its working life even though the enzyme remains active.
4. **Powder or solution**: Lyophilised powder stores longest and needs reconstitution and aliquoting. A ready-made solution is convenient and has a shorter working life. For occasional use, powder is the better buy; for routine use, a stabilised solution aliquoted once saves handling errors.
5. **Inactivate or remove before the next step**: Heat inactivation, a column clean-up or a phenol extraction all work and all have consequences for yield. Choose deliberately, because residual activity carried into an enzymatic step is a failure that presents as a reagent problem three steps later.

**What grade of Proteinase K do I need?**

Molecular biology grade, certified nuclease free, whenever the nucleic acid downstream must stay intact. General grade is fine for protein removal, tissue dissociation and any use where nucleic acid integrity is irrelevant.

**How should I dose Proteinase K?**

By activity units against the supplier's stated substrate rather than by mass, since specific activity varies by lot and preparation. Re-check when changing supplier or lot.

**How do I inactivate Proteinase K?**

Heat inactivation, a column clean-up or a phenol extraction, chosen for what the next step tolerates. It survives conditions that stop most enzymes, so removal has to be deliberate rather than assumed.

**Powder or ready-made solution?**

Powder for occasional use because it stores far longer; a stabilised solution for routine use because it removes a weighing and reconstitution step. Either way, aliquot rather than repeatedly thawing one stock.

## RNase A and ribonucleases

Ribonuclease is two things in one laboratory: a reagent you deliberately add to remove RNA from a plasmid preparation, and the contaminant that destroys every RNA experiment on the same bench. Both facts follow from the same property, which is that the enzyme is extraordinarily robust and survives autoclaving. This page covers buying it deliberately and controlling it thereafter.

1. **Choose the grade from what must survive**: RNase A for plasmid work must be certified DNase free, or it will nick the plasmid you are purifying. Read what the certification actually tested and at what sensitivity, because a nicking activity below a crude detection threshold is still enough to show on a careful gel.
2. **Solution or powder, and the boiling step**: Some preparations are supplied as solutions already treated to destroy residual DNase; others are powders that laboratories traditionally boil before use for the same reason. If you buy a certified solution, do not repeat the treatment, and if you buy powder, follow the supplier's instruction rather than a remembered protocol.
3. **It survives autoclaving**: Ribonuclease refolds after heating and is not removed by autoclaving, which is why contaminated glassware and plasticware stay contaminated. Decontamination needs a chemical treatment or disposable consumables, and this is the fact that governs how an RNA bench must be run.
4. **Separate the RNA work physically**: Dedicated pipettes, dedicated tips, dedicated reagents and ideally a separate bench are the standard control, because the enzyme travels on hands, dust and aerosols. A laboratory that does plasmid preparations and RNA work at the same bench will eventually lose an RNA experiment.
5. **Inhibitors are protection, not a cure**: Recombinant ribonuclease inhibitors protect RNA in a reaction and have their own conditions and limits, including sensitivity to reducing agent concentration and temperature. They do not substitute for keeping the enzyme off the bench in the first place.

**Does autoclaving destroy RNase?**

No. Ribonuclease refolds after heating and survives autoclaving, which is why contaminated labware stays contaminated and why RNA work needs chemical decontamination or dedicated disposables.

**What does a DNase free RNase A label promise?**

The preparation was tested for residual deoxyribonuclease activity and none was detected at the stated sensitivity. It matters for plasmid work, where residual activity would nick the DNA being purified.

**Do I still need to boil RNase A?**

Only if the supplier's instructions say so. Certified solutions are already treated, and repeating the treatment is unnecessary. Follow the data sheet for the product you actually bought.

**How do I stop RNase ruining RNA work?**

Physical separation: dedicated pipettes, tips, reagents and ideally a separate bench, plus gloves changed often. Inhibitors protect a reaction but do not make a contaminated bench safe.

## Thermal cyclers

A thermal cycler holds a temperature and changes it quickly, and almost every meaningful difference between instruments comes down to how uniformly it does the first and how fast it does the second. The block format then decides what consumables you are committed to for the instrument's life, which is the decision buyers most often make by accident. This page covers both, plus the verification that tells you a cycler is still doing what it claims.

1. **Block format commits you to consumables**: Whether the block takes individual tubes, strips, a plate of a given format, or is interchangeable, determines your plastic for years. Interchangeable blocks cost more and are worth it for a laboratory whose work varies; a fixed block is fine where the format is settled.
2. **Ramp rate, and whether you need it**: Faster ramping shortens runs substantially over many cycles, and it can change results for reactions sensitive to transition times. For routine end-point work the difference is time only; for anything being transferred between instruments, the ramp rate is part of the protocol.
3. **Gradient capability for optimisation**: A gradient block runs a range of annealing temperatures in one plate and turns a week of optimisation into one run. For a laboratory that develops assays it pays for itself quickly; for one that runs fixed validated protocols it is a feature that will not be used.
4. **Uniformity across the block**: The specification that matters is how much the temperature varies across the block, because a difference between wells appears as inconsistent amplification that looks like pipetting error. Ask for the uniformity figure and how it was measured, and verify it periodically.
5. **Lid design and evaporation**: A heated lid prevents condensation, and the pressure it applies has to suit the consumable. A lid that seals strips poorly loses volume over a long run, which concentrates the reaction and shifts results in a way that is hard to trace.

**What is the difference between a thermal cycler and a thermocycler?**

Nothing. The two names describe the same instrument and suppliers use both, sometimes on the same page.

**Do I need a gradient block?**

If you develop or optimize assays, yes, because it collapses a week of annealing temperature trials into a single run. If you run fixed validated protocols only, it is a feature that will sit unused.

**How important is ramp rate?**

It shortens runs noticeably across many cycles and can affect reactions sensitive to transition times. If a protocol will move between instruments, treat the ramp rate as part of the protocol rather than a convenience.

**Should a thermal cycler be calibrated?**

Yes, periodically and with an independent probe, because the instrument reports its own sensor rather than the reaction temperature. Uniformity across the block matters as much as absolute accuracy.

**Why do different wells give different results?**

Usually block uniformity or an unevenly heated lid. Ask for uniformity and ramp specifications, and verify uniformity on a used instrument before it joins a validated workflow.

**What decides a quantitative instrument's usefulness?**

The optics and the software: how many channels, how well they separate your dyes, the dynamic range, what the software exports and whether an audit trail exists. Price is the least informative specification.

## CDMO services

A contract development and manufacturing organization sells two quite different things under one name, and confusion between them is the most common source of disappointment in this category. Development is problem-solving work with uncertain duration; manufacture is scheduled capacity with a quality system attached. Organizations that are excellent at one are frequently ordinary at the other. This page covers how to tell them apart and what to establish before committing a program.

1. **Separate development from manufacture**: Ask which of the two the organization is genuinely built for, and look at where its senior technical people sit. A development-strong provider will discuss your molecule's problems; a manufacturing-strong one will discuss slots, changeover and batch records. Both are valuable and you probably need them at different times.
2. **Capacity and the slot**: Manufacturing capacity is booked ahead and slots are the real currency. Establish what is actually reserved, what happens if your program slips, and what happens if theirs does. A slot with no contractual consequence on either side is a statement of intent.
3. **Technology transfer is the risky interval**: Most failures happen when a process moves between sites. Ask how transfer is run, who leads it, what documentation accompanies it and how comparability is demonstrated at the receiving site. A provider with a written transfer protocol has done this before.
4. **The quality system and inspection history**: Ask which regulatory authorities have inspected the site, when, and the general outcome, and ask to see the site's approach to deviations and change control. A provider reluctant to discuss its inspection history is telling you something.
5. **Analytics: theirs, yours or both**: Decide whether methods transfer to the provider, stay with you, or are developed by them, and who owns them afterwards. Analytical method ownership is a recurring source of friction and is far easier to settle at contracting than in the middle of a campaign.

**What is a CDMO?**

A contract development and manufacturing organization: a provider that develops a manufacturing process for a product and makes material to it under a quality system. Development and manufacture are different capabilities and few providers are equally strong at both.

**What is the difference between a CDMO and a CRO?**

A contract research organization runs studies and generates data. A CDMO develops processes and manufactures material. Some organizations offer both, in which case ask which side your assigned team comes from.

**What should I check before technology transfer?**

That there is a written transfer protocol, a named lead, a documentation package, and an agreed way to demonstrate comparability at the receiving site. Transfer is where most outsourced processes go wrong.

**Who owns the analytical methods?**

Whatever the contract says, so settle it at signature. Method ownership, along with the cell bank and process improvements, is a recurring source of dispute once a program is under way.

**How do I choose a manufacturing partner?**

On fit rather than capacity: experience of your modality at your scale, the analytical methods already running there, regulatory history in your market, and whether your batch size suits them. Ask for a reference program and meet the people who would run yours.

**What usually delays a programme?**

Analytics. Transferring and qualifying release methods routinely takes longer than process development, so a schedule showing process work and no analytical development will slip.

## Custom peptide synthesis

Custom peptide synthesis is quoted per residue at a stated purity and scale, and the two specifications that move the price most are the ones buyers leave vague. Everything on this page concerns peptides supplied for laboratory research: reagents for assays, immunogens, standards and substrates. This site does not cover peptides intended for human use, and nothing here is guidance on administering anything to a person.

1. **Length and sequence difficulty**: Price rises with length and rises faster where the sequence is difficult: hydrophobic stretches, aggregation-prone regions and repeats all reduce coupling efficiency. Providers will often assess a sequence before quoting, and a sequence flagged as difficult is a warning about yield rather than a sales technique.
2. **Purity grade is the largest lever**: Crude material suits epitope mapping and screening; intermediate grades suit antibody production; the highest grades suit quantitative assays and structural work. Each step up costs a purification pass and a yield loss, so specify the grade the application needs rather than the highest available.
3. **Scale, and what you will actually use**: Peptides are usually quoted at a delivered mass, and laboratories routinely order far more than an assay will consume before the material degrades. Order what will be used within a sensible period and reorder, unless the peptide is for a long program.
4. **Modifications are priced individually**: Terminal capping, labels, biotinylation, phosphorylation, cyclisation, non-natural residues and isotope labels each add reagent cost and steps. List every modification in the request, because a modification added after quoting frequently changes the route.
5. **Analytics and what you receive**: Expect mass spectrometry confirming identity and a purity determination by chromatography with the trace. For quantitative work, ask about the net peptide content, since a mass of lyophilised material includes counterion and water and is not all peptide.

**What purity should I order for custom peptides?**

Crude for screening and epitope mapping, an intermediate grade for raising antibodies, and a high grade for quantitative assays and structural work. Each grade above the need adds a purification pass and cost without improving the result.

**What does net peptide content mean?**

The proportion of the delivered mass that is actually peptide, the rest being counterion, water and residual salts. It matters whenever a concentration has to be accurate, and it is reported separately from chromatographic purity.

**Why is my peptide insoluble?**

Sequence composition largely determines solubility, and hydrophobic or aggregation-prone peptides need a specific solvent approach. Ask the provider for a recommendation and test a small quantity before committing the batch.

**Are research peptides suitable for human use?**

No. They are supplied and labeled for laboratory research only. This site covers them as laboratory reagents and publishes no guidance on human use.

## Custom protein production

Production is the stage after a process exists: the question is no longer whether the protein can be made but how to get a defined quantity at a defined quality, repeatedly. That shifts every decision toward specification, release testing and comparability between lots, and away from the development questions that dominated earlier. This page covers how to commission a production batch and how to make the second one match the first.

1. **Size the batch from consumption and stability**: Order what will be used inside the material's demonstrated stability period plus a reserve, not the largest batch the unit price favors. Protein that degrades in storage is more expensive than a second batch, and stability is a property of your formulation rather than a general rule.
2. **Write a release specification**: Identity, purity by a stated method, concentration by a stated method, activity where relevant, aggregate, endotoxin and appearance, each with an acceptance limit. This document is what the provider manufactures against and what an acceptance dispute is settled by.
3. **Comparability between lots**: Agree what evidence shows a new lot is equivalent to the last: the same release panel plus a side-by-side test in your own assay against a retained reference. Without a reference aliquot you have no way to distinguish a lot difference from an experimental one.
4. **Formulation, storage and shipping**: The buffer, the concentration, the presence of carrier protein or stabiliser, and whether the material is frozen or lyophilised all affect how it behaves on arrival. Specify them, and require a temperature record with the shipment.
5. **Secure the inputs for the repeat**: The cell bank, the construct and the process description are what make the second batch possible. Establish who holds them, where the backup is, and what your rights are before the first batch rather than after.

**How large a batch of custom proteins should I order?**

Enough for consumption within the material's demonstrated stability period plus a reserve. A larger batch at a better unit price is a false economy if the protein degrades before it is used.

**What should a release specification contain?**

Identity, purity and concentration each by a stated method, activity where relevant, aggregate, endotoxin where the material touches cells, and appearance, each with an acceptance limit agreed before manufacture.

**How do I know a new lot matches the old one?**

The same release panel, plus a side-by-side comparison against a retained aliquot of the previous lot in your own assay. Without a retained reference the comparison cannot be made.

**What do I need to secure repeat supply?**

Access to the cell bank, the construct and a written process description, with a backup held somewhere other than the provider's freezer. Settle this before the first batch.

## Antibody cell line development

Developing a line to express an antibody is the general cell line problem with one specific complication: the product has quality attributes that vary between clones and that matter as much as how much protein each clone makes. A funnel that selects on titre alone will reliably find a high producer whose glycosylation or aggregation profile makes it unusable. This page covers how the screen should be built and what evidence the resulting line must carry.

1. **Screen on quality from the first cut**: Aggregation, charge variants, glycan profile and fragmentation vary between clones and are not correctable downstream without cost. Bringing analytical screening forward into the early funnel eliminates unusable clones before the expensive stages rather than after them.
2. **Titre in the right format**: Productivity measured in a small static format does not predict productivity in a fed-batch process, and clones reorder between the two. Screen in a format that resembles the intended process as early as the funnel allows, or the ranking you carry forward is the wrong ranking.
3. **Clonality evidence at deposition**: The line must be shown to derive from a single cell, with evidence captured at the time, typically imaging of the deposition event retained as a record. Regulators ask for this directly, and it cannot be reconstructed later.
4. **Stability over the required generations**: Expression and product quality are assessed over a generation number that covers the intended manufacturing scale with margin. A clone that drifts in glycan profile over the generations a campaign needs is a clone that has to be found now rather than at scale.
5. **Bank and document before you scale**: A research bank with identity, sterility, mycoplasma and adventitious agent testing, plus the process description, is what lets the program continue if anything changes. Establish where the backup sits and who may access it.

**Why screen antibody clones on product quality?**

Because aggregation, charge variants and glycan profile vary between clones, are expensive to correct downstream, and are not predicted by titre. A titre-only funnel reliably produces high producers that cannot be used.

**What clonality evidence is expected?**

Evidence captured at the time of single cell deposition, usually imaging retained as a record, showing the line derives from one cell. It is asked for directly and cannot be reconstructed afterwards.

**How many generations should cell line stability testing cover?**

Enough to cover the intended manufacturing scale with a margin, since expression and quality attributes can drift. A study that stops short of what the process needs will have to be repeated.

**Do host cell line licences affect my programme?**

Usually yes. Host lines and expression systems carry terms including milestones or royalties that attach to the program rather than the provider. Get the full position in writing before development starts.

## Custom cell culture media

Custom media is commissioned for one of three reasons: a catalog formulation does not support the cells well enough, a component has to be removed for regulatory or supply reasons, or the volume makes a bespoke blend cheaper than a catalog product. Only the first is a science question, and it is worth being clear which of the three is driving the project before starting. This page covers the decision, the format and the terms.

1. **Be clear which problem you are solving**: Performance, composition control and cost lead to different projects. A performance project needs a development program with screening; a composition project may be a straightforward removal or substitution; a cost project is a procurement exercise on a formulation you already have.
2. **Liquid or powder**: Liquid is convenient and heavy to ship and store, with a shorter shelf life. Powder stores far longer and needs water quality, mixing and filtration capability plus the validation that goes with them. Volume and in-house capability decide this rather than preference.
3. **Raw material sourcing and traceability**: Ask where components originate, whether any are animal derived, and what traceability the provider holds. For anything that may become a regulated process, animal origin components are a question that will be asked and is far easier to avoid than to justify.
4. **Formulation ownership**: Establish who owns the formulation and whether you can have it made elsewhere. A bespoke medium you do not own is a single-source dependency for the life of the process, which is a serious position for a program that succeeds.
5. **Lot control and bridging**: Agree lot sizes, retention samples and what happens when a component supplier changes. Media lot variation is a classic invisible cause of drifting culture performance, and it is only detectable if you kept a reference and recorded which lot produced which result.

**When are custom media for cell culture worth commissioning?**

When catalog formulations and a supplement screen genuinely fail to support the cells, when a component must be removed for regulatory or supply reasons, or when volume makes a bespoke blend cheaper. Exhaust the catalog first.

**Liquid or powder cell media?**

Liquid for convenience at modest volumes; powder for long storage and large volumes, provided you have the water quality, mixing and filtration capability and the validation to match.

**Who owns a custom formulation?**

Whatever the contract says. If you do not own it or cannot have it made elsewhere, the medium is a single-source dependency for the life of the process.

**How do I control media lot variation?**

Test each new lot against a retained reference in your own culture, keep retention samples, and record which lot produced which result. Lot variation is otherwise invisible and easily mistaken for an experimental effect.

**Should antibiotics be routine in culture medium?**

No. Routine use masks poor technique and low-level contamination, including mycoplasma, and selects for resistance. Use them for a defined purpose such as primary isolation or a selection, and work without them otherwise.

**Why did my cells change after a media reorder?**

Most often a formulation, serum lot or coating batch difference. Record the formulation, the serum lot and any coating with its batch, and change medium deliberately with an overlap rather than at the bottle.

## Compound microscopes

A compound microscope stacks two magnifying stages, an objective and an eyepiece, to reach magnifications a single lens cannot, and it is the instrument almost every laboratory means by microscope. Two quotations at similar prices can deliver very different instruments, and the difference is almost never in the headline magnification. This page covers what to check and in what order.

1. **Objective correction and numerical aperture**: Numerical aperture rather than magnification sets what the instrument can resolve, and the correction class sets how flat and color-true the field is. A quotation that lists magnifications without numerical apertures and correction classes is withholding the specification that decides image quality.
2. **The condenser is half the optical path**: An adjustable condenser with an iris, properly set up, is what produces contrast and resolution in transmitted light. Instruments sold with a fixed condenser are cheaper and cannot be aligned properly, and no objective compensates for that.
3. **Illumination quality and stability**: Even, adjustable illumination with a stable color temperature matters for imaging and for hours of comfortable viewing. Light emitting diode sources are now standard and vary considerably in evenness and in whether intensity changes shift color.
4. **Head, eyepieces and field number**: A binocular or trinocular head with adjustable interpupillary distance and dioptre correction is worth specifying, and the eyepiece field number determines how much of the image you actually see. A large field number with mediocre objectives shows more of a poorer image.
5. **Mechanical quality**: Stage movement, focus feel and absence of drift are what make an instrument pleasant over years, and they are the things a brief demonstration reveals immediately. Ask to operate it rather than to watch it operated.

**What is a light compound microscope?**

An instrument that magnifies in two stages, objective and eyepiece, to view thin specimens at high magnification, usually in transmitted light. It is the standard laboratory microscope for slides and sections.

**What magnification do I need?**

Ask about numerical aperture instead. Resolution is set by numerical aperture, and magnification beyond what it supports enlarges the image without revealing more detail.

**What is the difference between a compound and a stereo microscope?**

A compound instrument gives high magnification of thin transmitted-light specimens with shallow depth of field; a stereo instrument gives low magnification, long working distance and a three-dimensional view of solid objects.

**Does the condenser matter?**

A great deal. An adjustable condenser with an iris, correctly set up, produces the contrast and resolution the objectives are capable of. A fixed condenser cannot be aligned and limits the whole instrument.

## Dissecting microscopes

A dissecting or stereo microscope is bought for working distance and comfort rather than for resolution, because the job is manipulating something with tools while looking at it. That makes the specification list unusual: the space under the objective, the quality and direction of the light, and whether the head can be positioned over a large object matter more than optical performance. This page covers those in the order they bite.

1. **Working distance first**: The space between the objective and the specimen is where your hands and instruments go. Too little and the work is impossible whatever the optics; auxiliary lenses trade magnification for working distance and are worth specifying if the task needs room.
2. **Zoom range or fixed magnification**: A continuous zoom lets you find an object at low power and examine it at high, which is how dissection actually proceeds. Fixed magnification instruments are cheaper and sharper at their design points and suit a repetitive task at a known scale.
3. **Lighting is the biggest quality difference**: Incident light from above shows surface detail; transmitted light from below shows outline and internal structure in translucent specimens. A ring light is even and can flatten relief; a dual gooseneck gives directional shadows that reveal it. Many tasks want both available.
4. **Stand type**: A pillar stand suits small specimens on a base plate; a boom stand reaches over large objects, plates and equipment and is far more flexible. A boom costs more and is the right answer whenever the specimen will not fit under a pillar.
5. **Ergonomics for long sessions**: Eyepiece height and angle, interpupillary adjustment and dioptre correction decide whether somebody can dissect for hours. If the instrument will be used all day, this matters more than any optical specification on the quotation.

**What is a dissecting microscope used for?**

Working on a specimen with tools under magnification: dissection, sorting, microsurgery, electronics inspection and specimen preparation. It gives low magnification, long working distance and a three-dimensional view.

**How much working distance do I need?**

Enough for the specimen and your instruments together. Measure the tallest thing you will work on and add room for hands and tools, then check auxiliary lens options if the standard distance falls short.

**Ring light or gooseneck?**

A ring light gives even shadow-free illumination that can flatten surface relief. Goosenecks give directional light that reveals it. Many laboratories fit both and choose per specimen.

**Pillar or boom stand?**

A pillar stand suits small specimens on the base plate. A boom stand reaches over large objects, plates and equipment and is much more flexible, at a higher price and a larger footprint.

## Automated fluorescence microscopes

Automating a fluorescence microscope replaces a person moving a stage and refocusing with a system that does it reproducibly and overnight, and the value is entirely in reproducibility and unattended throughput. What it does not do is improve the image, which is still set by the objective, the filters, the light source and the camera. This page separates the automation decisions from the optical ones, because buyers frequently pay for the first and are disappointed by the second.

1. **Autofocus method decides unattended reliability**: Hardware autofocus tracks a reference surface and is fast and robust across a plate; software autofocus searches the image and is slower and can fail on sparse fields. For overnight or plate-scale acquisition, hardware focus is usually the difference between a complete run and a wasted night.
2. **Stage accuracy and repeatability**: For time-lapse and for revisiting positions, the stage must return to the same place repeatedly. Ask for the repeatability specification rather than the resolution, since the second is easy to quote and the first is what your experiment depends on.
3. **Light source and filters match your fluorophores**: Confirm the excitation and emission bands actually supplied against the dyes you use, including any planned multiplexing, and ask about crosstalk between channels. Generic filter sets that half-match a dye panel are the most common cause of disappointing multi-color data.
4. **Camera sensitivity against speed**: Faint signals need a sensitive camera with appropriate pixel size for the objective; fast live-cell work needs frame rate. These pull in opposite directions, so decide which experiments dominate rather than choosing a camera that is mediocre at both.
5. **Software, analysis and export**: Acquisition software drives the hardware; analysis may be in the same package or elsewhere. Confirm that raw images and their metadata export in an open format, because data locked in a vendor format is a problem that grows with every year of acquisition.

**What does an automatic microscope achieve over a manual one?**

Reproducible, unattended acquisition across many positions, wells or time points. It does not improve image quality, which is still determined by the objective, filters, light source and camera.

**Hardware or software autofocus for microscope automation?**

Hardware focus tracks a reference surface and is far more reliable for plate-scale and overnight acquisition. Software focus searches the image and can fail on sparse or low-contrast fields.

**How do I choose filter sets for fluorescence microscopes?**

From the dyes you actually use, including planned multiplexing, and ask about crosstalk between channels. Generic sets that approximately match a panel are a common cause of poor multi-color data.

**What about data storage?**

Plan it before the instrument arrives. Automated acquisition fills storage quickly, and the network between microscope and analysis becomes part of the workflow. Agree a retention policy at the same time.

## Light microscopes

A light microscope forms an image from visible light, and the illumination is not an accessory to that image but half of how it is formed. Laboratories that spend carefully on objectives and accept whatever lamp arrives are throwing away resolution they paid for. This page covers the optical choices and the lighting choices together, because in practice they are one decision.

1. **Transmitted or incident**: Transmitted light passes through a thin specimen and suits sections, smears and cultures. Incident light reflects off an opaque surface and suits materials, electronics and solid samples. Instruments are built for one and adapted to the other, so decide from the specimen.
2. **Set up Kohler illumination and keep it**: Aligning the field and condenser irises so the light source is evenly spread and the light cone matches the objective is what delivers the contrast and resolution the optics can produce. It takes minutes, it is taught rarely, and it improves most instruments immediately.
3. **Light emitting diode sources**: Modern sources run cool, last far longer than halogen and hold color temperature as intensity changes, which matters for imaging. Evenness across the field still varies between manufacturers, so look at a field of view rather than at a specification.
4. **Contrast methods extend what light can show**: Unstained transparent specimens show almost nothing in plain brightfield. Phase contrast, differential interference contrast and darkfield each reveal different features, and each needs matching condenser and objective components specified with the instrument.
5. **Filters and color**: A green interference filter sharpens detail for monochrome imaging, and neutral density filters let you reduce intensity without shifting color. These cost very little and are routinely missing from otherwise well specified instruments.

**What is a light microscope?**

An instrument that forms an image using visible light, as opposed to electrons or scanned probes. It covers compound, stereo and inverted instruments in transmitted or incident illumination.

**Why does illumination matter so much on a bright field microscope?**

Because it is part of how the image is formed. A properly aligned condenser and an even source deliver the resolution and contrast the objectives are capable of; a poor one wastes them.

**Are LED microscope lights better than halogen?**

Generally yes for stability, lamp life and color consistency as intensity changes. Evenness across the field still varies between makes, so look at an actual field of view before deciding.

**What is the resolution limit of a light microscope?**

It is set by the wavelength of light and the numerical aperture of the optics. Magnification beyond that limit enlarges the image without revealing further detail, which is called empty magnification.

## FTIR spectrometers

An infrared spectrometer is chosen by how the sample will be presented to it, because the sampling accessory determines what can be measured, how long it takes and how repeatable it is. Buyers compare resolution figures and signal to noise ratios and then discover the accessory they need is an extra purchase that changes the whole instrument choice. This page starts where the decision actually starts.

1. **The sampling accessory comes first**: Attenuated total reflectance suits solids, powders, pastes and liquids with almost no preparation and is why most laboratories now measure the way they do. Transmission cells, diffuse reflectance and specular reflectance each suit particular sample types. Choose the accessory from your samples, then choose an instrument that takes it well.
2. **Crystal material for attenuated total reflectance**: Diamond is chemically robust and handles hard samples; germanium and zinc selenide offer different refractive index and spectral range at lower cost and lower durability. The crystal decides both the usable range and how long the accessory survives your chemistry.
3. **Resolution beyond what you need costs time**: Condensed phase samples produce broad bands that a modest resolution resolves completely, while gas phase work genuinely needs finer resolution. Specifying high resolution for routine solid and liquid work lengthens every scan and improves nothing.
4. **Detector and beamsplitter set the range**: The standard room temperature detector covers the mid infrared adequately; a cooled detector gives far better sensitivity for small samples and microscopy at the cost of coolant handling. The beamsplitter defines the accessible range and is what limits extension into the far or near infrared.
5. **Purge, humidity and the environment**: Water vapour and carbon dioxide absorb strongly in the mid infrared and appear in spectra as interference. A purge or desiccation arrangement is part of the installation in a humid environment, and it is far easier to specify at purchase.

**What sampling accessory do I need for FTIR?**

Attenuated total reflectance for most solids, powders, pastes and liquids, because it needs almost no preparation. Transmission, diffuse reflectance and specular reflectance suit particular sample types and should be chosen from your samples.

**What resolution do I need?**

Condensed phase samples have broad bands that a modest resolution resolves fully. Gas phase work needs finer resolution. Over-specifying lengthens every scan without improving condensed phase spectra.

**Which ATR crystal should I choose?**

Diamond for chemical robustness and hard samples, and the alternatives where cost or a different spectral range matters. The crystal sets both the usable range and the accessory's working life against your chemistry.

**Is a library match an identification?**

No. It is a similarity score against whatever the library contains. Identity in a regulated setting requires a validated method against reference material.

**Infrared or Raman?**

Infrared for dry samples and where water is absent; Raman for aqueous samples, for measuring through transparent packaging and where no contact is possible. Raman costs more and struggles with fluorescent or colored samples.

## Particle size analysers

Two particle size instruments measuring the same material will report different numbers, and neither is wrong. Each technique measures a different physical property and reports an equivalent diameter derived from it, weighted differently, so a result is only meaningful alongside the technique and the dispersion conditions that produced it. This page covers choosing the technique first and the instrument second.

1. **Match the technique to the size range**: Dynamic light scattering covers the nanometre range and suits proteins, liposomes and colloids. Laser diffraction covers a broad range from sub-micron upward and suits powders, emulsions and suspensions. Imaging and sieving cover larger particles and give shape information. The range decides the technique.
2. **Understand what is being weighted**: Scattering techniques weight toward larger particles because scattered intensity rises steeply with size, so a small number of aggregates dominates a result. Number-weighted techniques such as imaging count each particle once. Converting between weightings introduces assumptions, so report what was measured.
3. **Dispersion decides reproducibility**: Whether particles are fully dispersed, and whether dispersion breaks them, is the largest source of variation in practice. Establish a dispersion method with its energy and duration, write it into the method, and keep it constant, because the sample preparation is the measurement.
4. **Wet or dry measurement**: Wet measurement disperses in a liquid and suits fine and cohesive powders; dry measurement is faster, uses no solvent and can attrit fragile particles. Running both on the same material and comparing is the quickest way to find out which one your sample tolerates.
5. **Optical properties for laser diffraction**: Diffraction analysis needs the refractive index and absorption of the material, and a wrong value shifts the reported distribution systematically. This input is frequently guessed, and it is worth establishing properly once for each material.

**Which particle analyzer technique should I use?**

Choose from the size range and what you need to know: dynamic light scattering for the nanometre range, laser diffraction for a broad range of powders and suspensions, imaging where shape matters, sieving for coarse material.

**Why does each particle size analyzer instrument give a different size?**

Because they measure different physical properties and report different equivalent diameters with different weightings. A result is only interpretable alongside the technique and the dispersion conditions.

**Wet or dry dispersion?**

Wet for fine and cohesive powders that need help separating; dry for speed and where no solvent is wanted, provided the particles survive the energy. Running both and comparing tells you which your material tolerates.

**Why does refractive index matter?**

Laser diffraction analysis uses the material's optical constants to convert the scattering pattern into a distribution. A wrong value shifts the whole result systematically, and it is often guessed rather than established.

**Why is a single viscosity figure not enough?**

Because apparent viscosity depends on shear rate, time and temperature for most real materials. A figure needs its geometry, shear rate and temperature; where the answer is a curve, a rheometer rather than a simple viscometer is the instrument.

**Why measure osmolality rather than calculate it?**

Because dissociation and interactions make the calculation unreliable, and it is the measured total particle concentration that determines tolerance by cells or by a patient. It is measured by freezing point depression or vapour pressure.

## Liquid chromatography

A liquid chromatography system is a pump, an injector, a column compartment, a detector and a data system, and buyers concentrate on the first four while the fifth is what they live with. The pressure class then determines which columns can be used and therefore what separations are available. This page covers the hardware decisions in the order they constrain each other, and the data system questions that matter more than they appear to.

1. **Pressure class decides your columns**: Conventional systems and the higher pressure class differ in the particle sizes they can run, and therefore in achievable speed and resolution. A method developed on one does not transfer unchanged to the other, so consider what methods must be run today and what must transfer in later.
2. **Detector follows the analyte**: Ultraviolet absorbance is the workhorse where the analyte has a chromophore. Diode array adds spectra and peak purity assessment. Refractive index handles analytes without a chromophore at the cost of gradient compatibility. Fluorescence gives sensitivity where the analyte allows. Mass spectrometry identifies as well as quantifies and is a different scale of purchase.
3. **Gradient formation and dwell volume**: High pressure and low pressure mixing differ in dwell volume, which shifts gradient methods between systems. If methods will move between instruments or come from elsewhere, the dwell volume is a transfer parameter and needs to be known rather than discovered.
4. **Injector, carryover and sample capacity**: Ask about carryover with your own difficult analyte rather than the specification figure, and check the autosampler's capacity, cooling and vial formats against how the laboratory actually queues work. Sample cooling matters for anything unstable at room temperature.
5. **The data system outlasts the hardware**: The chromatography data system holds raw data, processing methods and the audit trail, and it is where a regulated laboratory's integrity questions are answered. Choose it deliberately, and confirm how raw data is retained, who can reprocess and what the audit trail records.

**What pressure class of HPLC systems should I buy?**

It depends on the particle sizes your methods need. Higher pressure systems run finer particles for faster, sharper separations; conventional systems run established methods. Methods do not transfer unchanged between the two.

**Which of the liquid chromatography detectors should I choose?**

Ultraviolet or diode array for analytes with a chromophore, refractive index where there is none, fluorescence for sensitivity where applicable, and mass spectrometry where identification is needed. The analyte decides it.

**Why does dwell volume matter on a liquid chromatography machine?**

It sets how long a programmed gradient takes to reach the column, so a method developed on one system can behave differently on another. It is a transfer parameter and should be recorded with the method.

**How important is the data system?**

More than the hardware in the long run. It holds raw data, processing methods and the audit trail, and it is what a regulated laboratory is examined on. Choose it deliberately rather than accepting the bundled option.

**Why does my method fail when moved to mass detection?**

Usually the mobile phase. Non-volatile buffers and ion-pairing reagents foul the source, so volatile buffers have to be substituted, and that changes the separation. Plan for volatile conditions if the method will ever be coupled.

## Pipette calibration

A pipette calibration certificate is only useful if it reports numbers. Many do not: they record a pass, a date and a sticker, and a laboratory holding a folder of those cannot tell whether an instrument has been drifting for two years. This page covers what a calibration should measure, what the certificate must show, and how to decide the interval and the level of accreditation you actually need.

1. **Multiple volumes, multiple replicates**: Testing at several volumes across the instrument's range, with several replicates at each, is what reveals both systematic error and imprecision. A single volume tested once is a check that the instrument aspirates, not a calibration.
2. **As-found before as-left**: The as-found result is the one that matters, because it tells you how the instrument was performing during the work you already did. A service that adjusts first and reports only the final state has destroyed the information you were paying for.
3. **Stated acceptance criteria**: The certificate should state the limits applied and where they came from, whether the international standard's values or your own tighter internal specification. Limits that are not stated cannot be checked and cannot be compared between providers.
4. **Conditions and the balance**: Gravimetric calibration depends on temperature, humidity, barometric pressure and the balance used, and a credible certificate records them. This is also why a bench calibration in your own laboratory can be more representative than one performed at a service center.
5. **Accreditation, and when you need it**: An accredited certificate carries the weight of a third party assessment of the provider's competence and traceability. Regulated and accredited laboratories generally need it; a research laboratory frequently does not and can buy a competent unaccredited service far more cheaply.

**What should a pipette calibration certificate show?**

As-found and as-left results at several volumes with several replicates each, the mean and the variability at each volume, the acceptance limits applied and their source, and the ambient conditions and balance used.

**How often should pipettes be calibrated?**

Set the interval from use and from what depends on the result, with a formal external service on a fixed cycle and quick internal gravimetric checks between. Heavy or regulated use warrants a much shorter interval.

**Do I need an accredited calibration?**

Regulated and accredited laboratories generally do. A research laboratory often does not, and can buy a competent unaccredited service far more cheaply, provided the certificate still reports real numbers.

**Is on-site calibration better?**

It calibrates the instrument in the environment where it works and avoids long absences. Shipped service can cost less per instrument but needs spares to cover the gap, which is rarely priced in the quotation.

**Can we calibrate our own pipettes?**

You can verify them gravimetrically with a suitable balance, an evaporation trap and recorded conditions, against a written procedure with stated acceptance criteria. Where an accredited certificate is required, that is a different thing and has to come from an accredited provider.

**What should pipette calibration software record?**

Every weighing behind the result, not the result alone: the balance and its own calibration, the temperature, pressure and humidity used for the correction, each replicate at each volume, and who performed it. That is what makes a certificate reproducible rather than a number, and it is also what an assessor asks to see first.

## Sample management software

Sample management software answers one question well: where is this sample, and what has happened to it. It becomes a problem when it drifts into answering questions the laboratory information system already owns, at which point two systems disagree about the same sample and the laboratory trusts neither. This page covers what to expect from the category and where to draw the boundary.

1. **Model the storage as it really is**: Freezer, shelf, rack, box, position, and the same again for ambient and refrigerated storage, across buildings. The hierarchy must match the physical estate, because a model the laboratory has to translate in its head is a model people stop updating.
2. **Barcodes and scanning that survive the freezer**: Labels have to stay legible after cold storage and handling, and scanning has to work at the freezer door where the work happens. This is a consumable and hardware decision as much as a software one, and it is where implementations most often fail in practice.
3. **Capacity, occupancy and finding space**: Good systems tell you what is full, what is nearly full and where a new box can go. Laboratories run out of freezer space suddenly and expensively, and a system that surfaces occupancy early is worth real money.
4. **Chain of custody and audit trail**: Every movement, aliquot, thaw and disposal recorded against a person and a time. Where results support a regulated submission, the electronic record and signature obligations apply to this system as much as to any other.
5. **The boundary with the system of record**: Decide explicitly what sample management owns, which is location and custody, and what the laboratory information system owns, which is tests, results and release. Two systems that both claim to hold sample status will disagree, and reconciling them becomes somebody's permanent job.

**What does lab sample management software do?**

It records where every sample is, how it got there, what was taken from it and when it was disposed of, with an audit trail. Location and custody are its job.

**Are sample management systems the same as a LIMS?**

No. A LIMS owns tests, results, methods and release; sample management owns location and custody. Some LIMS include sample management, and where both exist the boundary must be decided explicitly.

**What is the hardest part of implementing lab sample tracking software?**

Registering what is already in the freezers, with labels that survive cold storage and scanning that works at the freezer door. Licensing is the small part of the project.

**Does it need an audit trail?**

Yes, and where results support a regulated submission the electronic records and signatures obligations apply to it as to any other system holding that data.

## Laboratory software

Laboratories buy overlapping software because vendors in each category extend into the others, and a laboratory that buys three products each claiming to do all five jobs ends up with three partial records of the same work. The way out is to decide which system is authoritative for which question before looking at any product. This page sets out the map.

1. **What arrived and where is it**: Sample registration, location and chain of custody. A laboratory information system or a dedicated sample management product owns this, and it is the foundation everything else hangs from, because a result that cannot be tied to a registered sample is not evidence of anything.
2. **What was done and what was found**: Tests against versioned methods, results, review and release. This is the laboratory information system's core and is the question a regulated laboratory is examined on most closely.
3. **What was tried and why**: The experimental narrative, the reasoning, the failures. An electronic lab notebook owns this, and a laboratory information system does it poorly because its data model is built around repeatable tests rather than around one-off experiments.
4. **Where the raw data lives**: Instrument files, images and large datasets with their metadata. Scientific data management systems and instrument data systems own this, and the system of record should hold a pointer and the reported result rather than trying to absorb the raw files.
5. **Who is doing what, when**: Scheduling, capacity and equipment booking. This is often a separate product or a module, and it is the one most safely bought independently because it touches the record least.

**What is the difference between a LIMS and an ELN?**

A LIMS is built around samples and repeatable tests with results and release; an electronic lab notebook is built around experiments and narrative. Testing laboratories usually need the first, discovery laboratories the second, and many eventually run both.

**In what order should laboratory software be bought?**

Sample registration and results first, because everything else references them. Then the notebook when narrative work justifies it, then a data management layer when raw file volume becomes a real problem.

**Should one vendor supply everything?**

Not necessarily. A single vendor reduces integration work and increases dependency, and suites are usually strong in one module and ordinary in the rest. Decide the boundaries first and then judge whether one vendor covers them well.

**Which laboratory software needs validating?**

Any system holding data that supports a regulated submission or product, which normally means the system of record and often the instrument data systems. The obligation follows the data, not the product category.

## HPLC columns

Column selection is where a liquid chromatography method is really decided, and it is also where methods quietly break, because two columns described identically by two manufacturers can separate differently. The bonded phase chemistry, the base silica and the end-capping all vary, and the catalog description conceals most of it. This page covers how to choose and how to protect a method from the column supply.

1. **Phase from the separation mechanism**: Reversed phase covers most small molecule work, with different ligand chemistries offering different selectivity for aromatic, polar and basic analytes. Normal phase, ion exchange and hydrophilic interaction each suit analytes that reversed phase handles poorly. Selectivity, not efficiency, is what usually solves a difficult separation.
2. **Particle size against system pressure**: Smaller particles sharpen peaks and raise back pressure, and the column must match what the system can deliver. Core shell particles offer much of the efficiency of very fine fully porous particles at lower pressure, which is often the practical answer on a conventional system.
3. **Pore size admits or excludes the analyte**: Small molecules suit the standard pore size; peptides and proteins need wider pores or they never reach the internal surface. A method transferred from small molecules to biomolecules without changing pore size is a method that will underperform for reasons that are hard to see.
4. **Dimensions set loading, solvent and time**: Length drives resolution and pressure; diameter drives loading and solvent consumption. Narrower columns save solvent and demand a system with low extra-column volume, otherwise the gains are lost in the tubing and the detector cell.
5. **Lot reproducibility and a second source**: Ask for the manufacturer's lot test chromatogram and acceptance criteria, and for a validated method keep enough of one lot to finish the work. Where a method must live for years, identify a second column that has been shown to work, before you need it.

**How do I choose an HPLC column?**

Start from the separation mechanism the analytes require, then set pore size to admit them, particle size from the pressure your system supports, and dimensions from loading and solvent considerations. Selectivity usually solves difficult separations rather than efficiency.

**Are columns from different manufacturers interchangeable?**

Not reliably. Base silica, bonded phase and end-capping differ even where descriptions match. Use a selectivity comparison tool to shortlist, then demonstrate equivalence on your own method before adopting.

**What are core shell HPLC columns for?**

They deliver much of the efficiency of very fine fully porous particles at lower back pressure, which makes them a practical way to sharpen separations on a conventional pressure system.

**How does a guard column in HPLC make a column last longer?**

Use a guard column, filter mobile phase and samples, stay inside the phase's pH and temperature limits, and follow the manufacturer's regeneration protocol. Track back pressure, which trends upward before a column fails.

**Can I substitute an equivalent column in a validated method?**

Not without data. Equivalence tables match published properties and cannot predict selectivity for your separation. Treat an equivalent as a candidate to evaluate, and where the method is registered, a change has to be justified.

**Why does my gradient method elute differently on another system?**

Usually dwell volume. Systems differ in the volume between the mixer and the column head, which shifts a gradient even when the arithmetic is right. Measure it on both systems and include it in the transfer.

## Test tube clamps

Three different tools are sold under similar names and used interchangeably to the detriment of everybody's glassware: the spring holder that grips a tube for heating, the clamp that fixes a tube to a stand, and the tongs that lift it. Each is right for one job and awkward in the others. This page separates them and covers the sizing and handling practice that keeps hot tubes intact.

1. **The spring holder is for heating in the hand**: A sprung wooden or metal holder grips a tube near the rim so it can be held in a flame at arm's length. It is for short, attended heating of a small tube and nothing else, and the tube is always pointed away from everybody present.
2. **The clamp is for fixing to a stand**: A clamp with adjustable jaws and a boss head fixes a tube to a retort stand for a long reaction, a distillation or an unattended cooling step. Jaw coating protects the glass, and the clamp should hold without being tightened until the tube complains.
3. **Tongs are for lifting and moving**: Tongs lift a tube out of a bath or a block and carry it a short distance. They are not for holding a tube steady during heating, because the grip depends on continuous hand pressure that is hard to maintain safely.
4. **Size the jaws to the tube**: Every holder and clamp grips a diameter range and fails at both ends. Measure the tubes in routine use rather than trusting a nominal size, since suppliers vary, and buy two sizes where the laboratory spans a wide range.
5. **Check the stand fit**: Clamps attach through a boss head to a rod of a particular diameter, and mismatched rods and bosses are a standing irritation in shared laboratories. Standardize the rod diameter across the laboratory and buy bosses to match.

**What is a test tube clamp used for, and how do test tube holders differ?**

Holding a tube in position on a stand, usually for an extended heating, reaction or cooling step. It is a different tool from the sprung holder used to hold a tube in a flame by hand and from tongs used to lift one.

**What size test tube clamp do I need?**

Measure the diameters of the tubes actually in use, since nominal volumes do not fix external dimensions and suppliers vary. Buy two sizes where the laboratory spans a wide range.

**Can I hold a tube in a flame with test tube tongs?**

It is not advisable. Tongs depend on continuous hand pressure and are meant for lifting and moving. Use a sprung holder for brief flame heating, or a clamp on a stand for anything longer.

**How should a tube be heated safely?**

Heat along the liquid rather than at the base, keep the tube moving, fill it only a fraction full, and point the open end away from everyone. Superheating can eject the contents forcefully.

## Laboratory supplies

Consumables are where a laboratory's money quietly goes, and they are also where the largest easy savings are, because the items are close to interchangeable at a given specification and most laboratories buy them without one. The order of operations matters: standardize the specification, then consolidate, then negotiate. Doing it the other way round buys cheaper items that do not work. This page covers the sequence and the exceptions.

1. **Write the specification before chasing price**: For each routine item, record what actually matters: material, sterility, certification, dimensions and any compatibility with equipment. Most laboratory arguments about a cheaper supplier are really arguments about an unwritten specification that the new item failed.
2. **Consolidate onto fewer suppliers**: Fewer suppliers means better pricing, fewer deliveries, fewer invoices and less administration, and it is usually worth more than the last increment of unit price. Consolidation also gives the laboratory leverage it does not have when spend is scattered.
3. **Keep a second source for what stops work**: For anything whose absence halts work, qualify a second product and keep it qualified. Supply disruption is a routine event rather than an emergency, and the laboratories that cope are the ones that qualified an alternative before they needed it.
4. **Hold stock where the lead time is long**: Consumables with long or unreliable lead times justify holding stock even at a storage cost. Consumables with short lead times do not, and holding them ties up money and expires on the shelf.
5. **Watch the specification creep**: Laboratories accumulate premium specifications that nobody chose deliberately: certified nuclease-free tubes for buffer work, the highest purity solvent for a wash step. A periodic review of what is actually required against what is actually ordered usually finds real money.

**How do I reduce laboratory supplies spend?**

Write specifications for routine items first, then consolidate onto fewer suppliers, then negotiate. Chasing price before specifying is how laboratories buy cheaper items that do not work.

**Should we use a single supplier?**

Consolidation is usually worth more than the last increment of unit price, but keep a qualified second source for anything whose absence stops work. Supply disruption is routine rather than exceptional.

**Can we switch a consumable used in a validated method?**

Not without assessing and documenting the change. Requalifying a method almost always costs more than the consumable saving, so validated items are the wrong place to look for savings.

**How much stock should we hold?**

Hold stock where lead times are long or unreliable and the item stops work. Do not hold stock of short lead time items, which ties up money and expires on the shelf.

## Type II restriction enzymes

BsmI is one of the restriction enzymes whose optimum incubation sits above the temperature most people set a digest at out of habit, and running it at the wrong temperature is the single most common reason a BsmI digest disappoints. Around that sit the usual questions of unit definition, buffer compatibility and methylation blocking, plus the useful option of substituting an isoschizomer. This page covers all of them.

1. **Use the supplier's stated temperature**: Several type II enzymes including this one have an elevated optimum, and a digest set at the conventional temperature will run slowly or incompletely. Read the data sheet rather than assuming, and check the block or bath actually holds the set point.
2. **Check thermal inactivation separately**: Whether an enzyme can be heat inactivated, and at what temperature, is a separate property from its incubation optimum. An enzyme that cannot be inactivated by heat needs a column clean-up or a spin before the next step, and that changes the protocol timing.
3. **Methylation can block the site**: Overlapping host methylation blocks some sites entirely, and whether yours is affected depends on the strain the DNA came from. Where it is, growing the plasmid in a methylation deficient strain is usually simpler than changing the strategy.
4. **Isoschizomers and neoschizomers**: Another enzyme recognising the same sequence may cut where this one cannot, or work in a buffer that suits your double digest better. A neoschizomer cuts the same site at a different position, which changes the overhang, so check the cut position and not only the recognition sequence.
5. **Confirm the site before you order**: Run the sequence against a restriction analysis tool and confirm the number and position of sites in the actual construct, including the vector. Ordering an enzyme for a site that the vector also carries is a wasted week.

**What temperature should a BsmI digest run at?**

Above the conventional digest temperature: check the supplier's data sheet for the stated optimum and set the block to it. Running it at the habitual temperature is the most common cause of an incomplete digest with this enzyme.

**What is an isoschizomer?**

A different enzyme recognising the same sequence. It may cut where the original is blocked by methylation or work in a more convenient buffer. A neoschizomer recognizes the same sequence but cuts at a different position, changing the overhang.

**Can BsmI be heat inactivated?**

Check the data sheet, because inactivation temperature is a separate property from incubation temperature and not every enzyme can be inactivated by heat. Where it cannot, clean up before the next enzymatic step.

**Why does my construct have unexpected sites?**

The vector usually carries sites too. Run a restriction analysis over the complete construct including the backbone before ordering, rather than over the insert alone.

**Blunt or sticky ends?**

Sticky ends ligate directionally with a matching overhang, which is what classical cloning depends on. Blunt ends ligate in either orientation and to anything, which is convenient and loses directionality.

**Which polymerase should I use?**

Choose on fidelity, processivity, strand displacement and template preference, because no enzyme is best on all four. Pick the one whose weakness does not matter to your application.

## Automated liquid handling

Automating liquid handling pays when the same transfers happen many times and the reproducibility matters, and it loses money when a laboratory buys throughput it does not have or a platform whose deck cannot hold the protocol. The instrument is rarely the limit; the deck layout, the tip strategy and the verification regime usually are. This page covers how to judge whether a protocol is worth automating and how to specify the platform that would do it.

1. **Count the transfers, not the plates**: The case for automation is built on how many individual transfers happen and how much variation between them costs. A protocol with few transfers and high judgement content is a poor candidate whatever the plate count, and a repetitive protocol with modest throughput can still be worth automating for consistency alone.
2. **Deck capacity is the real constraint**: Every plate, reservoir, tip box, module and waste position competes for deck space, and a protocol that does not fit on the deck needs an operator to intervene, which removes the walk-away benefit. Lay out the protocol on the candidate deck before the quotation, not after delivery.
3. **Air or positive displacement**: Air displacement channels handle aqueous liquids and are the norm. Viscous, volatile, dense or foaming liquids need positive displacement or acoustic transfer, and putting them on the wrong channel type produces a systematic volume error that no calibration corrects.
4. **Tip strategy drives running cost**: Disposable filtered tips remove carryover and are the recurring spend; washable fixed tips remove the spend and introduce a carryover question that has to be answered for every assay. Decide from the assay's sensitivity to carryover rather than from the tip price.
5. **Verify volumes independently**: The platform reports what it was told to dispense. Gravimetric or dye-based verification across channels and volumes is what tells you what it actually dispensed, and it should be scheduled rather than performed only when results look wrong.

**When are automated pipetting systems worth it?**

When the same transfers happen many times and variation between them costs something. Protocols with few transfers or high judgement content are poor candidates regardless of how many plates are involved.

**Do I need positive displacement channels?**

For viscous, volatile, dense or foaming liquids, yes, or the air cushion introduces a systematic error. Aqueous work is fine on standard air displacement channels.

**Disposable or washable tips on automated liquid handling workstations?**

Disposable filtered tips eliminate carryover and are the main recurring cost. Washable tips remove that cost and require carryover to be validated for each assay. Sensitivity to carryover decides it.

**How do I know the volumes are right?**

Verify independently and on a schedule, gravimetrically or with a dye-based method, across channels and across the volume range. The platform reports what it was instructed to do, not what it did.

## Liquid handlers

Liquid handlers fall into classes that are priced an order apart and are not substitutes: a small benchtop unit that replaces repetitive manual pipetting, a full deck platform that runs a whole protocol unattended, and specialist instruments such as acoustic dispensers that move very small volumes without contact. Buying across the class boundary is the expensive mistake in this category. This page sets out what each class is for.

1. **Benchtop, small deck**: A compact unit with a few positions that automates plate filling, serial dilution and reagent addition. It suits a laboratory replacing repetitive manual pipetting rather than one seeking walk-away operation, and it is affordable enough to sit beside a person who still runs the assay.
2. **Full deck platform**: A larger instrument with many positions, independent channels, a multi-channel head and room for modules. It suits a protocol that must run end to end unattended, and it needs deck layout, method development and maintenance to earn its price.
3. **Multi-channel heads**: A 96 or 384 channel head transfers a whole plate at once and is what makes plate-based screening fast. It cannot do the selective, well-by-well work that independent channels do, which is why many platforms carry both and why a head-only instrument is limiting.
4. **Acoustic and non-contact dispensing**: Acoustic transfer moves very small volumes with no tip at all, which removes carryover and tip cost and suits compound screening. The instruments are specialist and the liquid classes they handle are constrained, so they complement rather than replace a conventional handler.
5. **Channel independence and volume range**: Ask the range each channel covers accurately, not just its maximum, since accuracy at the bottom of the range is where assays are lost. A platform covering a wide range usually does so with more than one channel type.

**What is the difference between a liquid handler robot and a benchtop liquid handler?**

Benchtop units automate repetitive pipetting beside a person; full deck platforms run whole protocols unattended; acoustic and non-contact instruments move very small volumes without tips. They are priced an order apart and are not substitutes.

**Does a liquid handling station need a 96 or 384 channel head?**

If plate-to-plate transfers dominate, yes, because a head moves a whole plate at once. It cannot perform selective well-by-well work, so most useful platforms pair a head with independent channels.

**Can I use third party tips?**

Sometimes, but the tip affects volume performance and may affect support. If you switch, verify volumes gravimetrically with the new tip across the channels and the volume range.

**What limits an automated liquid handling robot in practice?**

Usually the deck, not the instrument. If a protocol does not fit on the deck, someone has to intervene and the walk-away benefit disappears. Lay the protocol out on the candidate deck before buying.

## Microplate readers

A plate reader is bought for the assays a laboratory runs today and lived with through the assays it runs in five years, which is why the mode list and the optical design matter more than the specification headline. The largest single decision is filters against a monochromator, and it is a genuine trade rather than a question of newer being better. This page covers the modes, the optics and the plate compatibility that catches people out.

1. **List the modes you genuinely need**: Absorbance covers colorimetric assays and nucleic acid quantitation; fluorescence intensity covers labeled assays; luminescence covers reporter and viability assays and needs a sensitive detector and good light exclusion. Time-resolved fluorescence and polarisation are specialist additions that should be bought for a named assay rather than for completeness.
2. **Filters or monochromator**: Filters transmit more light and give better sensitivity at their fixed wavelengths. A monochromator selects any wavelength and is far more flexible for method development, at a sensitivity cost. Laboratories running fixed validated assays usually want filters; laboratories developing assays usually want the monochromator, and hybrid instruments offer both.
3. **Plate formats and physical fit**: Confirm the reader takes every plate format the laboratory uses, including the deep well, low volume and 384 well plate types, and check plate height and skirt requirements. A reader that will not accept a plate the assay requires is a limitation nothing works around.
4. **Shaking, temperature and injectors**: Onboard shaking and incubation are needed for kinetic and cell-based work; reagent injectors are required for flash luminescence assays where signal decays immediately. Injectors add cleaning and maintenance, so buy them for an assay you actually run.
5. **Software, export and integration**: Check how raw data leaves the instrument, whether the analysis you need is included or extra, and whether the reader can be driven by an automation platform if that is the direction of travel. Data locked in a proprietary format is a growing problem.

**Filters or monochromator in a fluorometer plate reader or a multimode plate reader?**

Filters give better sensitivity at fixed wavelengths and suit fixed validated assays. A monochromator selects any wavelength and suits method development, with some sensitivity cost. Hybrid instruments carry both.

**Which detection modes do I need in a multimode microplate reader?**

Absorbance and fluorescence intensity cover most routine work; luminescence needs a sensitive detector and good light exclusion. Time-resolved fluorescence and polarisation should be bought for a named assay, not for completeness.

**Will a microtiter plate reader take 384 well plates?**

Check explicitly, along with deep well, low volume and any specialist format you use, including plate height and skirt requirements. Format compatibility is a hard limit.

**Do plate colour and material matter in a fluorescence plate reader?**

Considerably. Clear plates for absorbance, black to reduce crosstalk in fluorescence, white to maximize signal in luminescence. The plate is part of the measurement and a change of supplier is a method change.

## Flow cytometers

A flow cytometer is configured rather than simply bought: the number of lasers and detectors decides how many markers can be measured at once, and that decides which experiments the laboratory can do for the life of the instrument. Buyers who start from the instrument end up with a configuration that does not run their panel; buyers who start from the panel do not. This page starts from the panel.

1. **Design the panel before the instrument**: Write down the markers to be measured together now and plausibly in three years. That panel determines the lasers and detector configuration, and adding a laser later is expensive or impossible depending on the instrument. Every other decision follows from this one.
2. **Spectral or conventional detection**: Conventional cytometers assign one detector per fluorochrome and need careful compensation. Spectral instruments record a full emission signature and unmix afterwards, which handles larger panels and similar dyes better. Spectral costs more and changes the analysis workflow, so evaluate both with your own panel.
3. **Analyser or sorter**: An analyser measures and discards; a sorter physically separates populations for culture or sequencing. A sorter is substantially more expensive, needs more skill and space, and where the samples are infectious it requires aerosol containment. Many laboratories rightly use an analyser in house and a core facility for sorting.
4. **Containment for sorting**: Sorting generates aerosols, and with human or infectious material that means the sorter sits inside a containment enclosure with a defined aerosol management and testing regime. This is a facilities decision with a facilities budget, not an accessory.
5. **Sample handling and throughput**: Plate loaders, tube loaders and sample cooling decide whether a large experiment is practical. An instrument that can only run tubes by hand will limit throughput long before its optics do.

**How many lasers do I need on a flow cytometer?**

It follows from the panel. Write down the markers to be measured together now and in the foreseeable future, and configure the lasers and detectors to that panel, because adding a laser later is often impossible.

**Spectral or conventional flow cytometry?**

Spectral instruments record a full emission signature and handle large panels and spectrally similar dyes better, at higher cost and with a different analysis workflow. Conventional detection remains adequate for smaller panels.

**Analyser or FACS flow cytometry: do I need a sorter?**

Only if populations must be physically recovered for culture or sequencing. Sorters cost far more, need more skill, and with infectious or human material require aerosol containment. Many laboratories use an analyser in house and a core facility to sort.

**What containment does sorting need?**

Sorting generates aerosols, so with human or infectious samples the instrument sits inside a containment enclosure with a defined aerosol management and testing regime, which is a facilities project rather than an accessory.

**Flow cytometry vs FACS: what is the difference?**

FACS, fluorescence-activated cell sorting, is flow cytometry on a sorter: the cells are measured the same way and chosen populations are then physically recovered for culture or sequencing, where an analyser measures and discards them. A sorter costs far more, needs more skill, and with human or infectious material has to sit inside a containment enclosure because sorting generates aerosols.

## Laboratory inventory management

Inventory software fails for one reason above all others: people stop updating it, and an inventory nobody trusts is worse than a shelf somebody looks at. Every evaluation question that matters therefore comes back to whether the system makes recording easier than not recording. This page covers what the system has to track and the adoption questions that decide whether it survives its first year.

1. **Consumables, chemicals and reagents are different problems**: Consumables need stock levels and reorder points. Chemicals need location, quantity, hazard information and regulatory status. Biological reagents need lot, expiry and storage conditions. A system strong at one and weak at the others will be worked around for the rest.
2. **Scanning at the point of use**: Recording has to happen where the item is taken, on a device that is to hand, in seconds. Systems that require someone to return to a computer are systems whose data rots, and this single property predicts adoption better than any feature list.
3. **Expiry and reorder that reach a person**: Alerts have to arrive where people already look rather than inside an application nobody opens. Ask how notifications are delivered and whether they can be routed to a shared channel or mailbox.
4. **Ordering and the finance boundary**: Decide whether the system raises requisitions, holds catalogs and prices, and how it meets the organization's purchasing system. Integration here saves real effort and is also where implementations stall, so establish what is actually possible early.
5. **Hazard and regulatory data**: For chemicals, the system should hold or link safety data, flag controlled substances and support the reporting an inspector may ask for. Retrofitting that onto an inventory built only for stock levels is substantial work.

**What should lab inventory management software track?**

Consumable stock and reorder points, chemical location, quantity and hazard data, and reagent lot, expiry and storage conditions. These are three different problems and systems vary in how well they handle each.

**Why does a lab inventory management system fail?**

Because people stop updating them. Recording has to be faster than not recording, at the point of use, on a device to hand. Adoption predicts success far better than the feature list does.

**Is lab inventory software the same as sample management?**

No. Inventory tracks consumables, chemicals and reagents you buy; sample management tracks specimens you generate or receive, with chain of custody. Some products cover both and the boundary should be decided deliberately.

**How should a lab inventory system implementation start?**

With one store or freezer done properly rather than everything done partially. A system people find genuinely faster spreads on its own; one nobody trusts does not.

## Lab inventory management

Most writing about laboratory inventory is about software, and most laboratories with an inventory problem do not have a software problem. They have no stock policy, no named owner and no routine count, and adding a system to that produces an expensive version of the same disorder. This page is about the practice a system would support: what to decide, who decides it, and what the law already requires you to know about your chemicals.

1. **Set minimum and maximum levels per item**: For each routinely used item, agree the level at which you reorder and the level above which you are wasting money and shelf life. Without those two numbers, ordering is a series of individual judgements and the result is simultaneous stockouts and expiry.
2. **Give every category a named owner**: Somebody owns reagents, somebody owns consumables, somebody owns chemicals, and that person decides specification and level. Shared responsibility for stock reliably produces both duplicate purchases and empty shelves.
3. **Count something every week**: A rolling cycle count of a small part of the inventory, rather than a full annual count, keeps the record honest with very little effort and catches drift early. A full count once a year tells you how wrong you were, which is less useful.
4. **Keep a chemical register that satisfies an inspector**: You need to know what chemicals you hold, in what quantity, where, and what their hazards are, and that information has to be available to the people working with them. This is a standing duty rather than a good practice, and it is the part of inventory with legal weight.
5. **Deal with expiry deliberately**: Decide which items are used past their date with a documented justification and which are not, and record the decision rather than leaving it to whoever opens the fridge. Blanket rules in either direction waste money or invalidate results.

**Do I need free lab inventory software?**

Not first. Set minimum and maximum levels, name an owner per category and start a rolling count. Software supports a practice; it does not create one, and adding it to disorder produces an expensive version of the same disorder.

**What must a chemical inventory contain?**

What you hold, in what quantity, where it is, and what its hazards are, with that information available to the people who work with it. This is a standing duty rather than good practice.

**How often should we count stock?**

A small rolling count every week keeps the record honest with little effort and catches drift early. A single annual count mostly tells you how wrong the record had become.

**Can we use reagents past their expiry date?**

Sometimes, with a documented justification and appropriate verification, and never for work where the date forms part of the evidence. Decide and record the policy per item rather than leaving it to the bench.

## Reagent inventory

Reagents differ from other laboratory stock in one way that changes everything: results depend on the lot. A system that records how many vials remain but not which lot was used cannot answer the question that eventually gets asked, which is whether a change in results came from the reagent. This page covers the lot-level requirements that separate a reagent system from a stock counter.

1. **Track the lot, not just the item**: Every receipt is a lot with its own certificate, expiry and storage history, and every use should record which lot was consumed. This is the property that makes the inventory useful for investigation rather than only for reordering.
2. **Hold the certificate with the lot**: Certificates of analysis arrive as documents and are lost as documents. Storing them against the lot record means the acceptance criteria and the measured values are available when somebody questions a result rather than in an inbox nobody can search.
3. **Expiry, open-vial life and storage**: Many reagents have a shelf life once opened that is far shorter than the printed expiry, and a storage condition excursion can end a lot regardless of date. The system should record opening and support a storage exception rather than tracking the printed date alone.
4. **Reservation for a study**: A study running for months needs enough of one lot to finish, and the system should let that quantity be reserved so it is not consumed by routine work. Discovering mid-study that the lot is gone forces an unplanned bridging exercise.
5. **Bridging when a lot changes**: Where a lot change is unavoidable, the system should make it visible and support recording the bridging comparison against a retained reference. An undocumented lot change is the classic invisible confound in a long dataset.

**Why track reagents at lot level?**

Because results depend on the lot. An inventory that records quantity but not which lot was used cannot answer whether a change in results came from the reagent, which is the question that eventually gets asked.

**What about open-vial life?**

Many reagents have a much shorter life once opened than the printed expiry suggests. The system should record opening and support storage exceptions, not track the printed date alone.

**How do I handle a lot change mid-study?**

Reserve enough of one lot to finish the study where possible. Where a change is unavoidable, bridge the new lot against a retained aliquot of the old one in your own assay and record the comparison.

**Should reagent inventory integrate with results?**

Yes, through a shared lot identifier. That is what lets a question about a drifting assay be answered by looking rather than by recollection.

## LIMS instrument integration

Instrument integration is the part of a laboratory information system project that overruns, and it overruns because it is scoped as a line item when it is a series of small software projects. Each instrument is its own negotiation with its own output format, its own firmware and its own quirks. This page sets out the three levels of integration, what each costs in effort, and how to scope the work so the estimate survives contact with the instruments.

1. **Inventory the instruments properly**: Make, model, firmware version, output format, whether the output is a file, a serial stream or a database, and who supports it. This document is the actual scope of the integration project, and producing it before contracting is the single most useful thing a buyer can do.
2. **Level one: result capture**: The instrument produces a file and the system parses it into results against the right samples. This is the commonest and cheapest integration and delivers most of the value, since it removes transcription and the errors that come with it.
3. **Level two: worklist download**: The system sends the instrument its worklist, so sample identity flows outward as well as inward and the operator does not key sample names into the instrument. This removes the other half of the transcription risk and costs meaningfully more.
4. **Level three: full bidirectional control**: The system schedules and drives the instrument and receives status as well as results. It is appropriate for high-throughput automated environments and is rarely justified elsewhere, and it is where integration budgets disappear.
5. **Ask who owns the interface afterwards**: Firmware updates, system upgrades and instrument replacements all break interfaces. Establish who maintains each one, at what cost, and what the response is when an interface fails the week a report is due.

**What does LIMS instrument integration cost?**

It is priced per interface and depends on whether a supported driver exists for that make, model and firmware. Build the instrument inventory with firmware versions before contracting, because that document is the real scope.

**What are the levels of LIMS integration?**

Result capture, where the system parses instrument output; worklist download, where sample identity also flows outward; and full bidirectional control, where the system drives the instrument. Cost rises steeply across the three.

**Why does a LIMS interface to an instrument break?**

Instrument software updates change output formats, and system upgrades change the parsing side. Supported drivers are more robust than bespoke parsers, and every interface needs a repeatable test that can be re-run after any upgrade.

**What should an interface do with data it cannot parse?**

Fail visibly and record the rejection. A parser that silently skips what it does not understand produces a partial record that nobody notices, which is worse than an obvious failure.

## Automated sample storage

An automated store is bought for three things: retrieval speed at scale, a complete audit of every access, and protection of samples from the temperature excursions that manual picking causes. It is a large capital purchase with facilities implications, and the case for it rests on how often samples are actually retrieved rather than on how many are stored. This page covers when it is justified and what has to be planned around it.

1. **Count retrievals, not samples**: A large archive that is rarely touched is well served by manual freezers and good records. A smaller collection accessed constantly is where automation pays, because the value is in picking speed and in avoiding the excursions that repeated manual searching causes.
2. **Excursion protection is the quality argument**: Every manual pick warms the whole box while somebody reads labels. An automated store picks individual tubes at temperature, which for long-term collections is a genuine sample quality benefit rather than a convenience.
3. **Tube format is a long commitment**: Stores work with specific tube and rack formats, usually barcoded on the base. Converting an existing collection is a substantial project in itself and must be costed as part of the purchase rather than assumed.
4. **Facilities, power and heat**: These are large machines with power, floor loading, heat rejection and access requirements, and they need to be sited with the building rather than placed in a laboratory. Involve the facilities team before the specification is fixed.
5. **Plan for failure before it happens**: Decide what happens on a mechanical failure, a refrigeration failure and a power failure: how samples are recovered manually, how long the store holds temperature, what backup power exists and who is called. A store holding irreplaceable material without a written failure plan is a risk nobody has accepted deliberately.

**When is automated sample storage worth buying?**

When retrieval frequency is high and the collection is valuable enough that temperature excursions during manual picking matter. A large but rarely accessed archive is usually better served by conventional freezers and good records.

**What happens if the store fails?**

That must be decided and written down before purchase: how samples are recovered manually, how long temperature holds, what backup power exists and who is called. Ask vendors for the manual recovery procedure specifically.

**Can I use my existing tubes?**

Usually not without conversion. Stores work with specific tube and rack formats, normally base-barcoded, and converting an existing collection is a substantial project that must be costed with the purchase.

**What should I try first?**

Base-barcoded tubes, a proper location model, a handheld scanner and disciplined records solve many of the problems that prompt an automated store enquiry, for a small fraction of the cost.

## Pipette tips

A pipette tip looks like the most interchangeable item in the laboratory and is one of the least. The seal at the tip cone is part of the measuring system, so changing tip changes the delivered volume, and the certifications printed on the box mean specific and different things. This page covers what to specify, what the labels mean and how to switch supplier without quietly changing every result.

1. **Fit is part of the measurement**: The tip seals against the instrument's cone, and the quality of that seal affects the delivered volume and the imprecision. Any change of tip should be verified gravimetrically on the instruments concerned before it is adopted across the laboratory.
2. **Filter tips and when they are required**: A hydrophobic filter stops aerosol and liquid reaching the instrument, protecting the pipette from contamination and the sample from whatever is in the pipette. For amplification work, for anything infectious and for volatile liquids they are standard practice rather than an upgrade.
3. **Low retention surfaces**: A treated surface reduces the film left behind, which matters for viscous liquids, for detergent-containing buffers and for precious samples at small volumes. It costs more and makes no measurable difference to routine aqueous transfers.
4. **Read the certifications separately**: Sterile, nuclease free, endotoxin tested and free from specific inhibitors are separate claims tested separately. Buy the ones your assay needs and do not assume any of them follows from another; sterile does not mean nuclease free.
5. **Racked, refill or bulk**: Racked tips are convenient and generate the most plastic; refill systems reduce waste and cost with a handling step; bulk bags are cheapest and are only appropriate where tips are loaded by hand into reusable racks. This is a waste and cost decision rather than a quality one.

**Do pipette tips need to match the pipette brand?**

Not necessarily, but the seal at the tip cone affects delivered volume and precision, so any change must be verified gravimetrically on the instruments concerned before being adopted.

**When do I need a pipette tip filter?**

For amplification work, infectious material and volatile liquids, where aerosol must not reach the instrument or the sample. They also protect an expensive pipette from contamination that is hard to remove.

**What does low retention mean?**

A surface treatment that reduces the liquid film left in the tip, which matters for viscous liquids, detergent-containing buffers and small precious volumes. It adds little for routine aqueous transfers.

**Are sterile tips the same as RNase free pipette tips?**

No. Sterile, nuclease free and endotoxin tested are separate claims tested separately. Specify the ones your assay actually requires.

**Can I trust a pipette tip compatibility chart?**

Use it to find candidates, then verify gravimetrically on your own instruments at the volumes that matter. Fit tolerance sets the seal and varies between molds and production runs, so a chart is a shortlist rather than evidence.

## Bispecific antibodies

A bispecific antibody binds two different targets, and the interesting engineering problem is not the binding but the assembly: putting two different heavy chains and two different light chains into one molecule without producing a mixture of wrong pairings. Every format in use is an answer to that problem, and each answer costs something. This page covers the trade rather than the catalog.

1. **The chain pairing problem defines the field**: Co-expressing two heavy and two light chains produces many possible combinations, only one of which is the product you want. Engineering approaches steer heavy chain pairing and light chain pairing separately, and the efficiency of that steering determines how much of the harvest is product rather than by-product.
2. **IgG-like or fragment format**: IgG-like formats keep the constant region, and with it the long circulating half-life and the familiar purification and analytical tools. Fragment formats are smaller, penetrate tissue better and clear quickly, which for some purposes is the point and for others is a problem requiring a half-life extension.
3. **Geometry and valency matter functionally**: Where the two binding sites sit relative to each other, and how many of each there are, changes what the molecule does biologically. This is a design decision driven by the mechanism rather than by manufacturability, and the two frequently pull against each other.
4. **Developability, assessed early**: Aggregation propensity, thermal stability, expression titre, viscosity at concentration and the size of the by-product burden should be assessed while multiple candidates are live. Formats that look elegant on paper often screen out here, and finding that late is expensive.
5. **Analytics are harder than for a monoclonal**: Confirming that the molecule assembled correctly, and quantifying mispaired species, needs method development beyond a standard antibody panel. Budget analytical development explicitly, because it is routinely underestimated in bispecific programs.

**What is a bispecific monoclonal antibody?**

An antibody engineered to bind two different targets or two different epitopes. The engineering challenge is assembling two distinct heavy and light chain pairs into one molecule without a mixture of incorrect pairings.

**IgG-like or fragment format?**

IgG-like formats keep the long circulating half-life and familiar analytics; fragment formats are smaller with better tissue penetration and rapid clearance. The mechanism and the required exposure decide it.

**Why are analytics harder in bispecific antibody development?**

Because you must show the molecule assembled correctly and quantify mispaired species, which needs method development beyond a standard monoclonal antibody panel. Budget for it explicitly.

**Does format choice carry licence obligations?**

Usually yes. Pairing technologies and format platforms are patented and licensed with terms that attach to the program. Settle the position before a candidate is selected, since reversing a format choice is very costly.

## Monoclonal antibody production

Producing a monoclonal antibody is a well-trodden process, which means the questions are about scale, purity and documentation rather than about feasibility. The two decisions that set the price are whether expression is transient or from a stable line, and how far the purification goes. Both should follow from what the material is for. This page covers how to match the route to the use.

1. **Transient expression for speed and small scale**: Transfecting cells for a short production run gives material in weeks without developing a line. It suits research quantities, candidate comparison and early studies, and the cost per unit rises steeply as quantity grows.
2. **Stable line for quantity and repeatability**: Developing a stable line costs months and pays back at scale and where every batch must match. Programs that will need material repeatedly should start the line early, because it sits on the critical path of everything that follows.
3. **Purification depth follows the use**: Affinity capture alone suits many research uses. Polishing to remove aggregate, and additional steps to remove host cell protein and DNA, are needed where the material goes into cells, into animals or toward a regulated application. Each step costs yield as well as money.
4. **Endotoxin, aggregate and the formulation**: Specify an endotoxin limit for anything touching cells or animals, an aggregate limit, and the final buffer and concentration. Antibody delivered in the provider's default formulation frequently needs an exchange the customer then performs badly.
5. **Documentation matched to the destination**: Research material needs a certificate of analysis. Material heading toward a regulated application needs manufacturing records, a traceable cell bank and a quality system, and retrofitting those onto a research batch is not possible.

**Transient or stable expression for monoclonal antibodies production?**

Transient for speed and research quantities, since it needs no line development. Stable for larger quantities and batch-to-batch consistency, accepting several months of development first.

**How pure does the antibody need to be?**

Affinity capture alone suits many research uses. Cell-based work, animal work and anything heading toward a regulated application need polishing and host cell protein and DNA removal, each costing yield.

**What should antibody services show on a certificate of analysis?**

Identity, concentration and purity each by a stated method with the trace, aggregate content, and endotoxin where the material touches cells or animals, against limits agreed before manufacture.

**Can research material be used later for a regulated application?**

Generally not. Regulated use requires manufacturing records, a traceable cell bank and a quality system that cannot be applied retrospectively to a research batch.

## Protein production services

Once the expression system is decided, protein production quotes start to look interchangeable, and they are not. Providers differ in which systems they genuinely run at scale, what analytics are inside the price, how they handle a batch that misses specification, and what you own at the end. This page is about reading those differences out of quotes that look similar.

1. **Which systems does the provider actually run**: Most providers list every expression system and run one or two well. Ask how many projects in your system they completed in the past year and ask to speak to a customer. Capability breadth on a website is not capability.
2. **What analytics are inside the price**: Purity, concentration, identity, aggregate and endotoxin may each be included or extra, and a cheaper quote frequently excludes the analytics that make the material usable. Normalise every quote against the same analytical package.
3. **Timelines, and what they are measured from**: A stated turnaround may run from order, from receipt of materials, or from a successful feasibility stage. Establish the start point and what pauses the clock, because this is where expectations diverge most often.
4. **What happens when a batch misses specification**: Ask directly: who pays, who investigates, and is a repeat included. A provider with a clear written answer has been here before; one that treats the question as hypothetical will negotiate it while you are waiting for protein.
5. **Ownership and the repeat order**: Constructs, cell banks, process descriptions and any improvement made during the work should have their ownership settled in writing before the first batch. This is the leverage point, and it disappears once the project has started.

**How do I compare protein production quotes?**

Normalise them against the same specification and the same analytical package, establish what each timeline is measured from, and ask what happens when a batch misses specification. Quotes that look similar frequently include very different things.

**Does the provider need experience in my expression system?**

Yes, and websites list far more systems than providers run well. Ask how many projects in your system they completed recently and ask to speak to one of those customers.

**Should I do a small run first?**

Usually. A confirmation run verifies the process in that provider's hands with your materials, and it costs far less than losing a full-scale batch to something it would have exposed.

**What should I own at the end?**

The construct, the cell bank, the process description and any improvement made along the way, as agreed in writing before the first batch. After work starts, the leverage to settle it is gone.

## Recombinant protein expression

Most recombinant expression problems are one of three things: the protein is not made, it is made and insoluble, or it is made and inactive. Each has a different set of remedies and they are not interchangeable, so the first useful step is establishing which one you have. This page sets out that diagnosis and the variables worth changing, roughly in the order that yields results per unit of effort.

1. **Establish which failure you have**: Look for the protein in the soluble fraction, the insoluble fraction and the whole lysate. Absent everywhere means expression or stability; present only in the pellet means folding; present and soluble but inactive means folding, modification or a missing cofactor. The remedies differ completely.
2. **Construct boundaries are the highest-yield variable**: Truncating to a domain, moving the start and stop a few residues, and removing disordered termini frequently converts an unexpressable construct into a well-behaved one. Screening several boundary variants in parallel is more productive than optimizing one construct at length.
3. **Codon usage and the untranslated region**: Optimizing codons for the host and removing structure in the region around the start codon addresses translation initiation and elongation problems. This is cheap to try given synthetic genes and is worth doing early in a bacterial host.
4. **Lower the temperature and slow it down**: Reducing induction temperature and inducer concentration slows synthesis and frequently converts inclusion bodies into soluble protein. It costs a little yield and is the single most reliable solubility remedy in bacterial expression.
5. **Change host before you refold**: Refolding from inclusion bodies is laborious and often low yielding. Strains engineered for disulphide formation or rare codons, or a move to insect or mammalian expression, usually give better material for less work if the budget allows.

**Why is my recombinant protein not expressing?**

First establish whether it is absent, insoluble or inactive, by looking in the soluble fraction, the pellet and the whole lysate. The three have different remedies and treating one as another wastes months.

**What is the most effective change to try?**

Construct boundaries. Truncating to a domain or trimming disordered termini converts many unexpressable constructs into well-behaved ones, and screening several variants in parallel beats optimizing one at length.

**How do I get soluble membrane protein expression instead of inclusion bodies?**

Lower the induction temperature and inducer concentration first, then consider a solubility fusion partner or a different host. Refolding from inclusion bodies is laborious and usually a last resort.

**Should I use a solubility tag?**

Test it, and test whether the protein stays soluble after the tag is cleaved. A fusion that keeps an insoluble passenger in solution only while attached has not solved the problem.

## mRNA synthesis services

In vitro transcribed mRNA is made by a well-established route, and the differences between suppliers sit in three places: how the cap is added, whether modified nucleotides are used, and how thoroughly the double-stranded RNA by-product is removed. That last one determines how much innate immune response the material provokes, which is frequently the difference between a construct that works in cells and one that does not. This page covers the specification.

1. **Template and sequence elements**: The template's promoter, the untranslated regions and the poly-A tail all affect expression and stability, and the tail can be encoded in the template or added enzymatically with different consequences for length uniformity. Specify these rather than accepting a default.
2. **Capping route**: Co-transcriptional capping with a cap analogue is a single step; enzymatic capping after transcription is a second reaction with generally higher capping efficiency. Ask which is used and what capping efficiency is measured and reported, because uncapped RNA is both non-productive and immunostimulatory.
3. **Modified nucleotides**: Substituting modified uridine reduces innate immune activation and generally increases protein output in cells. Whether you want it depends on the application, since some experiments deliberately study that response. It changes the cost and should be specified explicitly.
4. **Double-stranded RNA removal**: Transcription generates double-stranded by-products that are potent innate immune triggers, and removing them is what separates good material from ordinary. Ask which purification is applied and whether residual double-stranded RNA is measured and reported per lot.
5. **Formulation and delivery**: Specify concentration, buffer and whether the material arrives naked or already formulated. Lipid nanoparticle formulation is a separate service with its own specification and its own analytics, and combining the two in one conversation obscures both.

**Why does double-stranded RNA content matter?**

Double-stranded by-products of transcription are potent innate immune triggers, so residual content largely determines how much response the material provokes in cells. Ask which purification is applied and whether residual content is measured per lot.

**Co-transcriptional or enzymatic capping?**

Co-transcriptional capping is one step; enzymatic capping is a second reaction with generally higher efficiency. Either way, ask what capping efficiency is measured and reported, since uncapped RNA is unproductive and immunostimulatory.

**Should I specify modified nucleotides?**

For most expression applications, modified uridine reduces innate activation and increases output. For experiments studying that response, unmodified material is the point. Specify it explicitly either way.

**What analytics should accompany the mRNA?**

Integrity and size distribution, capping efficiency, poly-A tail length, residual double-stranded RNA, residual template DNA and endotoxin, with the underlying traces rather than a summary.

## Antibody conjugation

Conjugation attaches a label or a payload to an antibody, and the two things that decide whether the conjugate works are the chemistry used and how many labels end up on each molecule. Too few and the signal is weak; too many and binding and solubility suffer. Most failed conjugations, though, fail before the chemistry starts, because the antibody arrived in a buffer that consumes the reagent. This page covers all three.

1. **Send the antibody in a compatible buffer**: Amine chemistry is consumed by any primary amine in the buffer, so material in a stabiliser containing one cannot be conjugated without a buffer exchange. Carrier protein in the formulation will also be labeled. Establish the buffer requirement before shipping, since this is the commonest avoidable failure.
2. **Choose the chemistry from what you can tolerate**: Amine chemistry is simple and labels randomly, including within binding regions. Thiol chemistry targets reduced disulphides and is more controlled. Site-specific approaches, whether enzymatic or through engineered residues, give homogeneous products and require more preparation.
3. **Set the labeling ratio deliberately**: The degree of labeling is a trade between signal and function, and the optimum differs by label and by application. Ask the provider to target a range and to report the measured ratio for the batch rather than the nominal one.
4. **Purify away free label**: Unreacted label produces background that looks like non-specific binding and is frequently mistaken for a bad antibody. Confirm that removal is part of the service and that residual free label is measured.
5. **Verify function after conjugation**: Conjugation can reduce affinity, especially with random chemistry at a high ratio. Test the conjugate in the intended application against the unconjugated antibody rather than assuming the label was neutral.

**Why did my antibody conjugation fail?**

Most often the buffer. Amine chemistry is consumed by primary amines in common stabilisers, and carrier protein in the formulation gets labeled too. Establish the buffer requirement and exchange before shipping.

**What degree of labelling should antibody fluorophore conjugation target?**

Enough for signal without impairing binding, which varies by label and application. Ask the provider to target a range and to report the measured ratio for your batch rather than a nominal figure.

**Random or site-specific conjugation?**

Random amine chemistry is simple and can label binding regions. Site-specific approaches give homogeneous conjugates with preserved binding and need more preparation and cost. Function sensitivity decides it.

**Should I test the conjugate?**

Yes, in the intended application against the unconjugated antibody, which means keeping an aliquot of the parent. Conjugation can reduce affinity and the loss is invisible without that comparison.

**How is antibody oligo conjugation different from a dye or enzyme?**

The payload is large, charged and made in a different chemistry, so the coupling is usually done in two steps through a linker rather than in one pot, and the product has to be separated from unreacted oligonucleotide by a method that can tell two large molecules apart. Expect a lower yield, a real purification step, and a characterisation that counts oligonucleotides per antibody rather than reading an absorbance ratio.

## Western blot loading controls

A loading control exists to show that differences between lanes are differences in your target rather than in how much was loaded, and it only does that job if the control protein is genuinely unaffected by the treatment under study. Several of the conventional choices are not, under conditions laboratories use constantly. This page covers choosing a control, proving it is invariant, and when to abandon housekeeping proteins entirely.

1. **Verify invariance in your own system**: The standard housekeeping proteins vary with cell cycle, confluence, hypoxia, differentiation and many treatments. Before adopting one, run your treatment series and confirm the candidate does not change. Inheriting a control from a published paper in a different system is how normalisation goes wrong quietly.
2. **Match the abundance to the target**: A control far more abundant than the target saturates at exposures where the target is barely visible, so the two cannot be quantified on the same blot. Choose a control of comparable abundance, or image the two separately with appropriate exposures.
3. **Check the molecular weight separation**: The control must resolve clearly from the target and from any non-specific band. A control running close to the target complicates stripping, reprobing and multiplexed detection.
4. **Stay inside the linear range**: Densitometry is only meaningful where signal is proportional to protein, and film and bright bands both saturate. Run a loading series once to establish the linear range for both target and control, and work inside it.
5. **Consider total protein normalisation**: Staining the membrane for total protein normalises to everything loaded rather than to one protein that may itself respond. It avoids the invariance problem entirely and has become the preferred approach where the target's abundance allows it.

**Which housekeeping gene should I use as a western blot control?**

One you have shown does not change under your treatment, at an abundance comparable to your target and resolving clearly from it. The conventional choices all vary under common conditions, so the validation matters more than the choice.

**Is total protein normalisation better?**

Often, because it normalises to everything loaded rather than to one protein that may itself respond to the treatment. It avoids the invariance problem and is generally preferred where the target's abundance allows.

**Why does a beta actin antibody band look identical in every lane?**

Frequently because it is saturated. A highly abundant control imaged at an exposure suited to a faint target will look uniform regardless of loading, which makes it useless as a control.

**How do I know I am in the linear range?**

Run a loading series once for both target and control and find where signal stops being proportional to protein. Quantify only inside that range.

## DNA extraction kits

Extraction kits are sold by sample type because the lysis step is where they differ, and the choice between column, magnetic bead and precipitation formats is really a choice about automation, throughput and how gently the DNA is handled. What almost never appears on the box is which downstream application the output suits, and that is the specification that matters. This page covers both.

1. **Start from the sample and its lysis**: Blood, cultured cells, tissue, fixed tissue, plants, soil and swabs each need a different lysis, and the rest of the kit is comparatively similar. Buying a kit designed for your sample type is mostly buying the right lysis chemistry, which is where yield is won or lost.
2. **Format follows throughput and automation**: Spin columns suit small numbers and need no equipment. Magnetic beads suit plates and automation and avoid the column's shear and clogging. Precipitation methods are cheap and laborious and can give very high molecular weight DNA. Choose from how many samples and whether a robot runs it.
3. **Yield or integrity, rarely both**: Long-read sequencing needs high molecular weight DNA and is damaged by vortexing, column passage and repeated pipetting. Short-read sequencing and amplification tolerate fragmentation. A kit optimized for yield may be the wrong kit for a long-read library, and the box will not say so.
4. **Read purity ratios for what they are**: Absorbance ratios indicate contamination classes rather than measuring inhibitors directly, and an acceptable ratio does not guarantee a clean amplification. Where inhibition is suspected, a dilution series or a spiked control answers the question that a ratio cannot.
5. **Check what the elution buffer does downstream**: Elution buffers containing chelators inhibit some enzymatic reactions, and buffer volume sets concentration. Confirm the elution composition suits the next step rather than discovering it during a failed reaction.

**How do I choose a DNA extraction and purification kit?**

From the sample type first, because lysis is where kits genuinely differ, then from throughput and automation, then from what the downstream application needs in terms of fragment length and purity.

**Column or magnetic bead extraction?**

Columns suit small numbers and need no equipment. Beads suit plates and automation and avoid shear and clogging, which also makes them gentler on high molecular weight DNA.

**Do purity ratios guarantee clean DNA?**

No. They indicate classes of contamination rather than measuring inhibitors. Where inhibition is suspected, run a dilution series or a spiked control instead of trusting a ratio.

**Which quantitation method should I use?**

A fluorescent dye-based assay specific for double-stranded DNA, because absorbance counts RNA and free nucleotides too and overstates concentration. For long-read work, size the extract as well as quantifying it.

**Do I need to enrich mitochondria?**

Only where copy number, heteroplasmy or a deletion is the question and the nuclear background would obscure it. Otherwise a total extraction with a mitochondrial target is simpler and loses less material.

**What decides a circulating DNA result?**

The pre-analytical handling: the tube, time to processing, centrifugation and freeze-thaw history. Leucocyte lysis swamps the short fragments, and no column chemistry recovers from that.

## Benchtop centrifuges

A benchtop centrifuge is the most shared instrument in most laboratories, which makes fit to the protocols in use more important than any single specification. The rotor range determines what the machine can do; the top speed on the datasheet is achievable only in the smallest rotor and is a poor comparison between models. This page covers specifying from the rotors, and the practical questions that decide whether the machine is pleasant to live with.

1. **List the vessels and the forces first**: Write down every tube, bottle and plate the machine must spin and the force each protocol requires. That list determines the rotor set, and the rotor set determines which machines are candidates. Working the other way round produces a machine that cannot run one important protocol.
2. **Top speed is a rotor property**: Advertised maximum speed usually applies to the smallest rotor. Ask for the maximum relative centrifugal force in each rotor you intend to buy, since that is what a protocol specifies and what limits the instrument in practice.
3. **Refrigeration if anything is temperature sensitive**: Enzymes, live cells, nucleic acids and long spins all need cooling, and frictional heating raises sample temperature noticeably during a long run even in a cool room. A refrigerated unit costs more to buy, run and site, and a laboratory that needs it and does not buy it will buy twice.
4. **Plate capacity and swing-out geometry**: If plates are spun, confirm the machine takes them, how many, and at what depth, including deep well and sealed plates. Plate spinning is a common late requirement and not every benchtop unit accommodates it.
5. **Noise, heat and bench footprint**: These machines are loud and vent heat, and a shared bench in a quiet laboratory is a poor location. Check the footprint with the lid open, since that is the space it actually needs.

**How do I choose a benchtop centrifuge?**

From the rotors. List every vessel and every force your protocols require, and let that determine the rotor set and therefore the candidate machines. Advertised top speed applies to the smallest rotor only.

**Do I need refrigeration?**

If you spin enzymes, live cells, nucleic acids or anything for an extended period, yes. Frictional heating raises sample temperature during long runs even in a cool room.

**Can a benchtop centrifuge work as a plate centrifuge?**

Many can with the right swing-out rotor, but not all, and deep well or sealed plates may not fit. Confirm plate capacity and depth explicitly if plates are in your protocols.

**When are sealed buckets required?**

Whenever a tube failure could release an infectious or toxic aerosol, including most work with human material. They also allow the rotor to be opened inside a biosafety cabinet.

**Why do two benchtop quotes differ so much?**

Because the rotor, buckets and adaptors are usually separate lines. Ask for the price configured against your own vessel list and the quotes generally converge; the one that stays low is normally missing a rotor.

## Choosing an enclosure

Three enclosures look similar from the corridor and protect entirely different things. A chemical fume hood protects the user from vapour and exhausts it. A biosafety cabinet protects the user, the work or both using filtered air. A laminar flow hood protects only the work and blows toward the operator. Choosing between them is a hazard assessment, not a preference, and getting it wrong is the most consequential equipment error a laboratory can make.

1. **Start from the hazard, not the catalog**: Chemical vapour, biological aerosol, particulate, radionuclide and combinations of these each point at a different enclosure, and some combinations need a specific type such as a ducted cabinet. Write down what is actually handled, in what quantity, before looking at any product.
2. **Chemical fume hood for chemical vapour**: A ducted chemical fume hood draws air away from the operator and exhausts outside, which handles volatile and toxic chemistry. It offers no product protection at all, so it cannot be used to keep a culture sterile.
3. **Biosafety cabinet for biological aerosol**: Cabinets are classed by what they protect and how they exhaust, and the class required follows from the agent and the containment level. Most recirculate through a filter, which means a cabinet is not a substitute for a fume hood when volatile chemicals are involved unless it is specifically a ducted type.
4. **Ductless enclosures within their limits**: Filtered ductless enclosures need no duct and suit a known, limited chemical inventory within the filter medium's capability, with a monitored change regime. They are unsuitable for unknown, varied or large-volume chemistry, and this limit is frequently exceeded in practice.
5. **Confirm containment as installed**: Whatever the type, containment is proven in the room it sits in with the air balance it will actually run under, not in the manufacturer's test rig. Establish who tests it, when, and what happens on a failure.

**What is the difference between a chemical fume hood and a biosafety cabinet?**

A fume hood draws chemical vapour away from the user and exhausts it outside, offering no product protection. A biosafety cabinet uses filtered air to protect the user, the work or both, and most recirculate, which makes them unsuitable for volatile chemistry unless ducted.

**Can I use a laminar flow hood for chemicals?**

No. A laminar flow hood blows filtered air across the work and toward the operator and exhausts into the room. It protects the work only and offers the user no protection whatsoever.

**Are ductless fume hoods acceptable?**

Within a known and limited chemical inventory that the filter medium retains, with a monitored change regime, yes. They are not suitable for unknown, varied or large-volume chemistry.

**Who decides which enclosure we need?**

The hazard assessment does, and where chemical and biological hazards combine it is worth asking a safety professional. This is the equipment decision with the largest consequences if it is wrong.

**Can one enclosure handle acids and biological work?**

Only if it is designed and certified for both, which is uncommon. The two protect different things in different directions, and a combination that is not designed for it protects neither the sample nor the operator.

## Water testing laboratories

Choosing a laboratory to send water samples to is a procurement decision with a compliance consequence, because a result from a laboratory outside its accredited scope may not be accepted by whoever asked for the test. The questions that matter are about scope, methods and reporting limits rather than price, and they are quick to ask. This page is about organizational testing, not consumer home test kits.

1. **Certification and the accredited scope**: Ask which program certifies the laboratory, for which analytes, by which methods, and ask for the current scope document. A laboratory may be entirely competent and still outside scope for your particular analyte, and the certificate is what a regulator reads.
2. **Methods by designation**: Regulatory testing specifies methods by published designation, and results are only comparable when the method is the same. Ask which method will be used for each analyte rather than accepting a general capability statement.
3. **Reporting limits in your matrix**: The limit the laboratory can report to depends on the matrix as well as the method, and a limit above your compliance threshold makes the result useless. Confirm the achievable reporting limit for your sample type against the threshold you must demonstrate.
4. **Sampling, containers and hold times**: Many analytes require specific containers, preservatives, filling technique and a hold time measured in hours. The laboratory should supply containers and written instructions, and a sample that misses a hold time is not salvageable however good the laboratory is.
5. **Chain of custody and turnaround**: For anything with a compliance or legal consequence, the chain of custody form is part of the result. Establish the turnaround for standard and expedited work and what happens if a sample fails a quality check on arrival.

**How do I check a water testing lab's accreditation?**

Ask for the current scope document naming the analytes and methods it is certified for, and confirm your analyte and method appear on it. Competence and accredited scope are not the same thing.

**Why does the method designation matter?**

Regulatory testing specifies methods by published designation, and results from different methods are not necessarily comparable. Ask which method will be used for each analyte.

**What are hold times?**

The maximum period between sampling and analysis for a given analyte, often only hours, alongside container and preservative requirements. A sample analyzed outside its hold time cannot be rescued.

**What should a water test report contain?**

The method, the reporting limit, the result with units, sampling and receipt dates and times, and the batch quality control outcomes with any affected results flagged.

## PCR reagents

Amplification failures are usually blamed on the template and are frequently the reagents, because polymerase choice, nucleotide quality and buffer composition each change what a reaction can do. Picking reagents deliberately, rather than using whatever is in the freezer, removes a large share of troubleshooting. This page covers the polymerase decision, the nucleotides and whether a master mix is worth the premium.

1. **Fidelity or robustness in the polymerase**: High fidelity enzymes with proofreading are required for cloning, sequencing libraries and anything where an error is carried forward. Robust, non-proofreading enzymes tolerate inhibitors and awkward templates and suit diagnostics and screening where an occasional error is irrelevant. The choice is the application's, not the catalog's.
2. **Hot start, almost always**: A hot start enzyme is inactive until the first denaturation, which suppresses primer dimers and non-specific products formed while reactions are set up at room temperature. The premium is small and the improvement in specificity is reliable.
3. **Nucleotide quality and storage**: The nucleotide mix is a small part of the cost and a common cause of quiet failure, because repeated freeze and thaw cycles degrade it. Buy a quality-tested mix, aliquot it once on arrival, and suspect it early when a reaction that used to work stops working.
4. **Master mix or components**: A master mix reduces pipetting steps and therefore variation, and it is usually the better buy for routine and high-throughput work. Separate components allow magnesium, nucleotide and enzyme concentrations to be tuned, which matters during method development.
5. **Primers and the design step**: No reagent recovers a poorly designed primer pair. Check specificity against the genome rather than against the target alone, check for dimers and hairpins, and match melting temperatures. This costs minutes and prevents most non-specific products.

**Which PCR polymerase should I use?**

A high fidelity proofreading enzyme for cloning, sequencing libraries and anything where errors are carried forward; a robust non-proofreading enzyme for diagnostics and screening where inhibitor tolerance matters more than fidelity.

**Do dNTPs go bad?**

They degrade with repeated freeze and thaw cycles, which is a common and quiet cause of a reaction that used to work failing. Aliquot on arrival and suspect the nucleotide mix early.

**Is a master mix worth the premium?**

For routine and high-throughput work, usually, because fewer pipetting steps means less variation. Separate components are better during development, when magnesium and nucleotide concentrations need tuning.

**Why use a hot start enzyme?**

Because it stays inactive until the first denaturation, suppressing primer dimers and non-specific products that form while reactions are assembled at room temperature. The specificity gain is reliable and the cost is small.

**Which pcr probes does the chemistry allow?**

A dual labeled hydrolysis probe is the default and needs a polymerase with strong five prime exonuclease activity, which rules out some high fidelity enzymes. A molecular beacon or a displacement design tolerates enzymes that a hydrolysis probe does not, and an intercalating dye needs no probe at all but cannot multiplex. Match the probe to the enzyme before ordering both, because a mismatch shows up as an amplification curve with no signal.

## Exome sequencing services

Exome sequencing quotes look interchangeable because they are all priced per sample, and the number that decides whether the data answers your question is not in the price. Capture kit, target definition, depth, coverage uniformity and what analysis is included vary widely between providers at similar prices. This page sets out what to specify and what to insist the provider reports back.

1. **The capture kit defines the exome**: Different capture designs target different regions, so two providers can both deliver an exome and cover different parts of your gene of interest. Ask which kit is used and check your genes of interest against its target definition before ordering rather than after.
2. **Depth is a mean, uniformity is the truth**: A mean depth figure conceals whether coverage is even. Ask for the proportion of the target covered above your required threshold, which is the specification that determines whether a variant can be called in the regions you care about.
3. **Input quality and quantity**: Providers state DNA input requirements and quality thresholds, and fixed tissue, degraded material and low-input samples each need a different library route. Declare the sample type honestly at quoting rather than letting it fail at the laboratory.
4. **What analysis is included**: Establish whether the price covers raw reads, alignment, variant calling, annotation or interpretation, and in which file formats. Bioinformatics is where quotes diverge most, and a cheap sequencing price with no analysis is not cheaper once somebody has to do the analysis.
5. **Data delivery and retention**: Agree how raw and processed data is delivered, how long the provider retains it, and what it costs to have it again. Sequencing data is large, and a provider that deletes after a short window is a risk if your storage plan is not ready.

**What should I specify for exome sequencing?**

The capture kit and its target definition checked against your genes of interest, the depth and the proportion of target covered above your threshold, the sample type and input, what analysis is included, and how data is delivered and retained.

**Is mean depth a useful specification?**

Only partly. A mean conceals uneven coverage, so ask instead for the proportion of the target covered above the threshold you need, which is what determines whether variants can be called where it matters.

**Exome or whole genome sequencing?**

Exome covers coding regions deeply and cheaply but misses non-coding, regulatory and structural variation entirely. Where the question could involve those, a genome at lower depth often answers more for similar spend.

**What analysis is normally included?**

It varies widely, from raw reads only to full annotation and interpretation. Establish the scope and the file formats explicitly, because bioinformatics is where apparently similar quotes differ most.

## AAV production

Adeno-associated virus production is quoted on titre, and titre is the specification most likely to mislead, because it can be measured by different methods giving different numbers and because a preparation's total particles include empty capsids that deliver nothing. Two lots at the same stated titre can behave very differently. This page covers what to specify so the material you receive is the material you priced.

1. **Serotype and the packaging limit**: Serotype determines tropism and is chosen from the target tissue and route. The packaging capacity is a hard limit on the genome, and a construct over it produces truncated genomes rather than a warning, so confirm the total size including the inverted terminal repeats and regulatory elements early.
2. **Titre is method dependent**: Genome-containing particles measured by amplification, total capsids measured immunologically and infectious units measured in cells are three different numbers. Ask which method is used, on which standard, and prefer providers who report more than one so the preparation can be understood.
3. **Empty and full capsid ratio**: Empty capsids carry no genome, contribute to the immune load and dilute the effective dose. The full to empty ratio should be measured and reported per lot, and a preparation with a poor ratio needs a higher total dose to achieve the same delivery.
4. **Purification route**: Density gradient separation enriches full capsids and is slower; chromatography scales better and varies in how well it separates empties. Ask which route is used and what ratio it achieves, since this is the largest quality difference between providers.
5. **Residuals and endotoxin**: Residual host cell protein and DNA, residual plasmid and helper sequences, and endotoxin should all have stated limits and lot results, particularly for in vivo work where they drive the response as much as the vector does.

**Why does the empty capsid ratio matter?**

Empty capsids carry no genome, so they dilute the effective dose and add to the immune load without delivering anything. The full to empty ratio should be measured and reported for every lot.

**Which titre method should I ask for?**

Ask which is used, because genome-containing particles, total capsids and infectious units are three different numbers. Providers reporting more than one give you a preparation you can actually understand.

**What limits construct size in AAV packaging services?**

The packaging capacity of the capsid, counting the whole genome including the inverted terminal repeats and regulatory elements. Exceeding it produces truncated genomes rather than an error, so confirm the total early.

**Can research grade AAV support a regulated application later?**

No. Material for regulated use needs a different process, documentation and testing package, and it cannot be applied retrospectively. Ask what would change before the program commits.

**Why do two titres for the same preparation differ so much?**

Because they measure different things: genome-containing particles, total capsids and infectious units differ by orders of magnitude. A titre without its method is not comparable, and dosing on the wrong one makes an in vivo result uninterpretable.

**Does the empty to full ratio matter?**

Yes, for potency and for antigen load: empty capsids contribute immunogenicity without expression. It is a specification, and so is aggregation, which raises apparent particle count and lowers infectivity.

## Cell line engineering

Cutting a genome is now routine. Proving that the line in your incubator carries the edit you intended, only that edit, and a phenotype caused by it rather than by clonal variation is where the work and the cost actually sit. Most disappointing edited lines are not failed edits; they are unvalidated ones. This page sets out what to specify and what to demand back, whether the work is done in house or bought.

1. **Decide whether you need a knockout or a knockdown**: A cut removes the gene product permanently and risks selecting for cells that tolerate its loss. Interference silences transcription reversibly and keeps the locus intact, which suits essential genes and dose response questions. Choosing by habit rather than by biology is the most common design error in this category.
2. **Design the edit around how you will detect it**: Choose guides where the expected outcome is detectable by a simple assay, and plan the genotyping before the transfection. An edit you cannot cleanly read is an edit you cannot report, and an out of frame indel is not proof that the protein is gone.
3. **Deliver by the route the cells tolerate**: Plasmid transfection suits easily transfected lines and leaves no integration when the nuclease is transient. Ribonucleoprotein delivery gives a short exposure window and lower off target risk. Lentiviral delivery is the route for difficult primary cells and for library scale work, and it integrates, which is a consequence to accept deliberately.
4. **Isolate clones and prove single cell origin**: Single cell cloning, documented at deposition, is the only basis on which a clonal genotype means anything. Keep the imaging or the dilution record, because a partner or a journal will ask.
5. **Confirm at the locus and at the protein**: Sequence the edited region deeply enough to see all alleles, and demonstrate loss or reduction of the protein itself. Sequence alone can miss an in frame product that still functions, and a blot alone cannot tell you what happened at the locus.
6. **Control for clonal variation before believing a phenotype**: Two or more independent edited clones and at least one clone that went through the whole process without an edit. A phenotype present in one clone and absent in the other is a clone, not a gene.

**How many validated clones are enough for KO cell lines?**

At least two independent edited clones plus a mock edited control that went through the same single cell bottleneck. One clone is an anecdote, because single cell cloning itself changes cells in ways unrelated to the edit.

**Does a knockout cell line need off target analysis?**

Proportionate to the use. For a discovery experiment with independent clones and a rescue, a guide design with a low predicted risk is often enough. For anything moving toward a product, targeted or unbiased off target assessment becomes an expectation rather than a nicety.

**When is a pooled screen the wrong tool?**

When the phenotype is not selectable or sortable. Pooled screens read survival or a sortable marker; anything requiring per well imaging, secretion measurement or a slow readout belongs in an arrayed format, which costs far more and should be budgeted honestly.

**Is an edited line still a custom cell line development project?**

Yes. It needs banking, identity testing, mycoplasma testing and stability like any other line, plus the edit evidence. Teams that treat editing as a bench technique rather than a line development project usually redo the characterisation later.

## Bioprocess development

Process development produces a process, but it also produces a set of constraints that will still be binding at commercial scale. The culture mode decides the media bill. The choice of single use or stainless decides the facility. The filtration train decides buffer volumes and the impurity profile. This page is about which decisions to make early and deliberately, and which can safely be deferred.

1. **Fix the culture mode against the product's stability**: Fed batch is simpler, better understood and adequate for stable molecules. Perfusion keeps residence time short, which protects labile products and raises volumetric productivity, at the cost of far more media, more complex control and a harder technology transfer. Choose from the molecule, not from fashion.
2. **Decide single use or stainless as a facility decision**: Disposable assemblies remove cleaning validation and shorten changeover, which suits multiproduct facilities and clinical scale. Stainless wins on consumable cost at large scale and sustained campaigns. The break even depends on batch size and campaign frequency, so model your own numbers rather than trusting a generic curve.
3. **Design the filtration train around buffer volume**: Concentration and buffer exchange dominate buffer consumption, tank capacity and processing time. Sizing the membrane area and the number of exchange volumes early tells you whether the facility can support the process, and it frequently sends people back to an earlier step.
4. **Watch shear and product quality across the train**: Pumps, valves and recirculation through a membrane all impose shear, and labile products aggregate. Screen the sensitive steps at small scale with the actual product rather than a surrogate, because aggregation discovered at pilot scale is expensive.
5. **Qualify the disposables as materials, not as hardware**: Extractables and leachables, film compatibility, gamma dose and supply continuity are material questions attached to every disposable component. A single sole sourced component can stop a campaign, so map the supply chain while the process is still flexible.
6. **Define the process description that will transfer**: Critical parameters with ranges, in process controls, hold times and acceptance criteria, written so a receiving site can run it. A process that exists only as the habits of the people who developed it does not transfer, whatever the batch records say.

**When do single use systems in bioprocessing stop being cheaper?**

At large batch sizes with sustained campaigns, where the recurring consumable cost outweighs the cleaning and changeover savings. At clinical scale, in multiproduct facilities and where changeover speed matters, disposables usually win on total cost.

**How is membrane area sized on TFF equipment?**

From the volume to process, the permeate flux the product allows and the time available, then confirmed experimentally because flux depends on the feed and fouls over time. Sizing from a supplier's nominal flux alone routinely underestimates the area required.

**Is perfusion worth the complexity in upstream bioprocessing?**

For unstable products and for intensified processes with high volumetric demand, yes. For stable molecules at moderate demand it adds media cost, control complexity and transfer risk without a proportionate return.

**What should be locked before a clinical campaign?**

The culture mode, the purification sequence, the critical parameters and the analytical methods. Changing any of those later triggers a comparability exercise, which costs more than getting them settled while the process is still cheap to change.

## Custom RNA synthesis

Custom RNA is made two ways and the boundary between them is a length. Short RNA is assembled chemically, base by base, which allows any modification anywhere and becomes progressively harder and more expensive as the molecule lengthens. Longer RNA is transcribed enzymatically, which is cheaper per base and constrains what modifications are possible. Knowing which route your order falls into explains most of the quote. This page covers both.

1. **Chemical synthesis for short RNA**: Solid phase synthesis gives complete control over sequence and modification position and is the route for guide RNA, small interfering RNA, antisense oligonucleotides and probes. Yield falls with each coupling, so cost per base rises with length and long chemical RNA becomes impractical.
2. **Transcription for long RNA**: Enzymatic transcription from a template produces long RNA economically and restricts modifications to those the polymerase accepts, applied throughout rather than at chosen positions. Messenger RNA and long non-coding RNA come this way.
3. **Modifications are priced individually**: Backbone changes, sugar modifications, terminal caps, dyes, quenchers and conjugation handles each add a reagent and a step. List every one at quoting, since adding a modification afterwards can change the synthesis route entirely.
4. **Purity grade follows the application**: Desalted material suits many routine uses; chromatographic or gel purification is needed for long constructs, for anything going into cells and for quantitative work. Each purification step costs yield, so specify what the application needs rather than the highest grade offered.
5. **Scale and the second order**: Scale is quoted as a delivered quantity, and RNA degrades in storage, so ordering far more than will be used is a poor trade. Where the program will re-order, agree the repeat price at the first order.

**Chemical synthesis or in vitro transcription for custom RNA?**

Chemical synthesis for short RNA where modification position matters, and transcription for long RNA where economy matters. The length of the molecule effectively decides it, and the route determines which modifications are available.

**What purity grade should I order?**

Desalted for routine short RNA, chromatographic or gel purification for long constructs, cell work and quantitative applications. Each step costs yield, so buy the grade the application needs.

**How should custom RNA be stored?**

Aliquoted on arrival and frozen, in nuclease-free conditions, with repeated freeze and thaw cycles avoided. Ribonuclease is ubiquitous and survives autoclaving, so dedicated consumables matter.

**What analytics should accompany the order?**

Mass confirmation for short RNA, integrity and size distribution for long RNA, purity with its method and trace, concentration, and for transcribed material residual template DNA.

## Cell-free protein synthesis

Cell-free synthesis makes protein in a reaction rather than in a living organism, which removes the constraints a cell imposes: the protein can be toxic, the reaction is open to additives, and results arrive in hours rather than days. It also removes what a cell provides, which is scale and the full machinery for folding and modification. This page covers where the trade is favorable and where expression in cells remains the better answer.

1. **Toxic and inhibitory proteins**: A protein that kills the host cannot be made in one, and in an open reaction toxicity is irrelevant because nothing has to survive. This is the clearest case for cell-free and it is why the route exists in many laboratories at all.
2. **Speed for screening and variants**: From template to protein in hours means variant panels, boundary screens and rapid checks become practical in a way that cell-based expression does not allow. Screening in cell-free and then producing the winner in cells is a common and sensible pattern.
3. **Non-natural amino acids and labeling**: The reaction is open, so unusual amino acids, isotope labels and precise labeling schemes can be supplied directly rather than engineered into a host. This is a genuine capability advantage for structural and chemical biology work.
4. **Membrane proteins with a supplied environment**: Detergents, nanodiscs or liposomes can be present as the protein is made, which helps membrane proteins that aggregate when expressed in cells. It is not a guarantee, but it is an option cells do not offer.
5. **Scale and cost are the limits**: Cell-free reactions are expensive per unit of protein compared with fermentation, and scaling them is a different exercise from growing more cells. For quantity, cell-based expression remains the economical route.

**When is cell-free protein synthesis the right choice?**

For toxic proteins, for rapid screening of many variants, where non-natural amino acids or precise labeling are needed, and for membrane proteins that can be given a lipid or detergent environment as they are made.

**Why not use cell-free for everything?**

Cost per unit of protein is far higher than fermentation and scaling is a different problem from growing more cells. For quantity, expression in cells remains the economical route.

**Which cell-free system should I use?**

Bacterial lysates for yield and cost with no eukaryotic modification; wheat germ or insect systems for complex eukaryotic folding at lower yield; reconstituted systems for the cleanest background at the highest price.

**Does cell-free protein fold correctly?**

Not automatically. Folding must be confirmed by an activity or structural assessment, exactly as with any other expression route. Presence on a gel is not evidence of function.

## Transfection reagents

Transfection efficiency is a property of the cell type at least as much as of the reagent, and the reagent that a colleague swears by may do nothing in your line. That is why every supplier offers a selection guide and why optimisation, done systematically once, is worth far more than switching products repeatedly. This page covers choosing a starting point and optimizing efficiently.

1. **Cell type comes first**: Easy adherent lines transfect with almost anything; primary cells, suspension cells and hard lines frequently need a specific chemistry or a physical method. Start from published or supplier data for your exact cell type rather than from a general recommendation.
2. **The cargo changes the chemistry**: Plasmid DNA, small interfering RNA, messenger RNA and ribonucleoprotein complexes have different sizes and charges and are not delivered equally well by one reagent. Suppliers sell different products for these because the difference is real.
3. **Optimize ratio and amount together**: Reagent to nucleic acid ratio, total nucleic acid per well and cell confluence at transfection interact, so vary them as a small grid once rather than one at a time repeatedly. A single properly designed optimisation plate saves weeks.
4. **Watch cytotoxicity, not only efficiency**: High efficiency with substantial cell death gives a population selected by survival, which distorts whatever you measure next. Track viability alongside efficiency and choose a condition that balances both.
5. **Serum, antibiotics and medium**: Some reagents are inhibited by serum and some require its absence during complex formation, and antibiotics in the medium raise toxicity in permeabilised cells. Follow the reagent's protocol on these points before concluding that it does not work.

**Which transfection reagent should I use?**

Start from data for your exact cell type and cargo rather than a general recommendation, because efficiency depends on the cell line at least as much as on the reagent. Suppliers publish cell type guides for this reason.

**Why is my transfection efficiency low?**

Common causes are the wrong chemistry for the cell type, poor nucleic acid quality, confluence outside the optimal window, serum or antibiotic interference, and an unoptimised reagent to nucleic acid ratio.

**Chemical reagent or electroporation?**

Chemical reagents are cheap and effective for many adherent lines. Electroporation and nucleofection reach primary and suspension cells that chemical methods cannot, at the cost of an instrument and some cell death.

**What controls should I include?**

A fluorescent reporter to measure efficiency, a non-targeting cargo to separate the cargo's effect from the procedure's, and untransfected cells. Record passage number and confluence alongside.

## Clinical trial management systems

A clinical trial management system runs the operational side of a study: sites, subjects, visits, monitoring, payments and milestones. It is not the system that collects the clinical data, and confusing the two is how organizations buy overlapping products and end up reconciling them by hand. This page sets out the boundary and what to establish during a selection.

1. **Operations here, clinical data elsewhere**: The trial management system tracks whether a visit happened; the electronic data capture platform holds what was recorded at it. Products in both categories extend toward each other, so decide which one is authoritative for each fact before any demonstration.
2. **Site and subject tracking at the level you work**: Confirm the system models your study structure, including multi-country studies, cohorts and amendments, without workarounds. Systems that need a workaround for the study shape you actually run will need one on every study.
3. **Monitoring, payments and milestones**: Visit planning, monitoring reports, site payments against milestones and budget tracking are where the operational value sits. Ask to see a monitoring workflow and a site payment run with realistic data rather than a dashboard.
4. **Integrations that exist today**: Ask what the system integrates with in production at a named customer: the data capture platform, the safety database, the trial master file, finance. Roadmap integrations are not integrations, and the boundaries are where the manual work accumulates.
5. **Validation and records**: Where the system holds records supporting a regulated submission, electronic record and signature obligations apply, along with validation of the configured system. Ask what the vendor provides toward that and what remains yours.

**What is a clinical trial management system?**

The operational system of record for a study: sites, subjects, visits, monitoring, payments, milestones and budgets. It tracks whether things happened, rather than holding the clinical data recorded at them.

**Is it the same as EDC clinical trial software?**

No. Data capture holds the clinical data collected at visits; trial management holds the operational picture. Products in both categories extend toward each other, so the authoritative source for each fact must be decided before selection.

**Does the system need validating?**

Where it holds records supporting a regulated submission, yes, along with the electronic records and signatures obligations. Establish what the vendor supplies toward validation and what remains the sponsor's.

**What should I check when clinical trial vendors demonstrate?**

A monitoring workflow and a site payment run using a study structure like your own, plus integrations that exist in production at a named customer rather than on a roadmap.

## Microcentrifuges

The microcentrifuge is the most used instrument in a molecular laboratory and the one bought with the least thought, usually on price and top speed. Neither is the right specification: protocols are written in relative centrifugal force, and what actually constrains a laboratory is which tubes fit and whether the machine can be cooled. This page covers the specification that matters and the accessories worth ordering with it.

1. **Work in force, not revolutions**: A protocol specifies relative centrifugal force because that is what acts on the sample, and the same rotational speed gives different force in rotors of different radius. Check the maximum force of the rotor you will actually use rather than the headline speed.
2. **Tube fit and adapters**: Standard microcentrifuge tubes are the base case, and strips, small tubes and larger tubes need adapters that may or may not exist for a given rotor. List every tube format the laboratory uses and confirm each before ordering.
3. **Refrigeration, and whether you need it**: Nucleic acid preparations, enzyme work and anything spun for an extended period benefit from cooling, and frictional heating is significant in a small high-speed rotor. Refrigerated units cost considerably more and take more bench space, so decide from the protocols rather than by default.
4. **Sealed rotors for hazardous samples**: Where human, infectious or toxic material is spun, a sealed rotor lid keeps an aerosol contained if a tube fails and allows the rotor to be opened in a cabinet. Confirm the sealed option exists for the rotor you are buying.
5. **Noise, footprint and where it sits**: These machines live on a crowded bench and are used constantly, so noise level, footprint with the lid open and lid action all matter more than they appear to. A machine nobody wants to sit beside gets moved somewhere inconvenient.

**How fast does a microcentrifuge need to be?**

Check the relative centrifugal force of the rotor you will use rather than the advertised speed, because force is what protocols specify and it depends on rotor radius as well as speed.

**Do I need a refrigerated microcentrifuge?**

For nucleic acid preparations, enzyme work and long spins, yes, since frictional heating in a small high-speed rotor is significant. For routine short spins at ambient it is an unnecessary cost.

**Will my tube strips fit?**

Only with the right adapter, and adapters do not exist for every rotor. List every tube format in use and confirm each one before ordering rather than after.

**When do I need a sealed rotor?**

Whenever a tube failure could release an infectious or toxic aerosol, including most human sample work. A sealed lid also lets the rotor be opened inside a biosafety cabinet.

**Are microcentrifuge tubes interchangeable between brands?**

Broadly, but not exactly. Dimensions and lid design vary enough that a rotor can hold one brand well and another loosely, and lids differ in whether they pop when heated. Check the fit in your own rotor before a large order.

## Autoclaves

An autoclave is a pressure vessel with a heat cycle, and the only thing that matters about it is whether the load reached the required condition for the required time. A machine that runs a cycle and reports success proves that the chamber reached temperature, not that the middle of your load did. This page covers sizing, cycle selection and the monitoring that turns a cycle into evidence.

1. **Size from loads, not from volume**: Chamber capacity matters less than whether your actual baskets, bottles and bags fit and can be loaded so steam reaches everything. Measure the largest routine item and the typical load, and ask about shelf configurations before choosing a size.
2. **Cycles for what you actually run**: Wrapped goods, liquids, waste bags and porous loads need different cycles, and liquid cycles need controlled cooling to prevent boiling over. A machine without a proper liquid cycle will lose media, and one without a waste cycle will not reliably decontaminate bags.
3. **Steam supply and siting**: Self-contained units generate their own steam and need water of a suitable quality; plumbed units need a building steam supply. Both need drainage, ventilation and a route for the heat and vapour they release, and they are heavy.
4. **Prove the load, not the chamber**: Chemical indicators show that conditions were met at their location, and biological indicators show that a resistant organism was killed. Place indicators inside the load, in its hardest to reach point, rather than relying on the cycle printout alone.
5. **Record keeping and pressure vessel duties**: Cycle records are the evidence of decontamination and should be retained. The vessel itself carries inspection and maintenance obligations, and those are the facilities team's concern as much as the laboratory's.

**What size laboratory autoclave do I need?**

Size from your actual loads rather than chamber volume: measure the largest routine item and the typical load, and check shelf configurations allow steam to reach everything.

**How do I know a load was sterilised?**

Place chemical indicators inside the load at its hardest to reach point, and use biological indicators periodically. The cycle printout shows what the chamber did, not what the center of the load experienced.

**Why do I need a separate liquid cycle?**

Liquids need controlled cooling at the end of the cycle to prevent boiling over when pressure is released. A machine without one will lose media and can be dangerous to open.

**What must never be autoclaved?**

Volatile solvents, corrosives, bleach and sealed containers, among others. Post a clear list at the machine, because most autoclave incidents come from something that should not have been in it.

**How big a chamber do I need?**

Size from the loads you actually run, not from a volume figure. Measure the largest item, count the trays in a typical load, and ask for the usable dimensions between shelves; nearly every complaint about capacity is about shape rather than litres.

**When is a pass through autoclave required?**

When the room layout has to separate contaminated material from clean, so that a load cannot travel back into the space it came from. It is a containment decision, and it needs the wall and the commissioning that go with it.

**Is a production machine better for a laboratory?**

Only if the load is repetitive and continuous. It buys throughput and duty-cycle reliability and costs flexibility, plus a steam and drainage requirement a laboratory machine avoids.

## Digital microscopes

A digital microscope replaces the eyepiece with a sensor and a screen, and it is bought for three real reasons: several people can see the same image at once, the instrument is comfortable to use for long periods, and measurement and documentation come out of the same workflow as viewing. What it does not change is that the optics in front of the sensor still determine the image. This page covers the trade.

1. **The optics still decide the image**: A high resolution sensor behind modest optics records a modest image at high pixel count, which is empty magnification with extra storage cost. Ask about the objective's numerical aperture and correction, exactly as you would for a conventional instrument.
2. **Where the screen genuinely helps**: Shared viewing during teaching, inspection and demonstration, comfortable posture for long sessions, and documentation captured as part of the work rather than as an extra step. These are real operational gains and they are the honest case for the format.
3. **Depth of field and stacking**: For solid, irregular specimens the shallow depth of field is the main frustration, and software that captures a stack and combines the sharp parts solves it. Where the work is three-dimensional objects, this feature is the reason to buy rather than a bonus.
4. **Measurement needs calibration**: On-screen measurement is only as good as the calibration behind it, per objective and per zoom position. Ask how calibration is performed and recorded, and verify with a stage micrometer before trusting any dimension the software reports.
5. **Export, storage and the software's lifetime**: Confirm images and measurements export in open formats with their metadata, and ask how long the software is supported. A digital instrument becomes unusable when its software stops running long before its optics wear out.

**Is a digital microscope with screen as good as a conventional one?**

The optics decide the image in both cases. A digital instrument wins on shared viewing, comfort and documentation; a conventional instrument with good eyepieces remains sharper and more responsive for fine work at high magnification.

**Does a higher megapixel count mean a better image?**

No. A high resolution sensor behind modest optics records a modest image at a high pixel count. Ask about numerical aperture and objective correction, as with any microscope.

**Can I measure with a digital microscope?**

Yes, provided the software is calibrated per objective and per zoom position and the calibration is verified with a stage micrometer. Uncalibrated on-screen measurement is not measurement.

**What is focus stacking for?**

It combines the sharp parts of images taken at different focal planes into one, which addresses the shallow depth of field that makes solid irregular specimens frustrating to image.

## Laboratory relocation

Relocating a laboratory is a project in which the moving is the easy part. What decides whether it goes well is decontamination, the cold chain for irreplaceable samples, the order in which things are disconnected and reconnected, and the requalification that instruments need before anyone can trust them again. This page sets out the sequence and the decisions inside it.

1. **Inventory and triage before anything moves**: A move is the best opportunity a laboratory ever gets to dispose of equipment it does not use and chemicals it cannot identify. Do that first, because paying to move and reinstall an instrument nobody uses is pure waste, and unknown chemicals are far harder to dispose of at the other end.
2. **Decontamination is a document as well as a process**: Every instrument needs decontaminating and a signed certificate stating what it was exposed to and how it was treated. Movers and service engineers are entitled to refuse equipment without one, and rightly, so this is on the critical path rather than beside it.
3. **Plan the cold chain separately and first**: Freezer contents are usually the most valuable and least replaceable thing in the building. Decide whether samples move in dry ice, in a validated shipper or in a powered vehicle, who accompanies them, and what the contingency is. Never let cold storage be part of the general move plan.
4. **Disconnect and reconnect in the right order**: Gas, water, vacuum, extract and specialist services need qualified disconnection and reconnection, and some instruments need manufacturer involvement to be moved without voiding support. Establish which ones early, because their engineer availability will set your dates.
5. **Requalify before you trust anything**: Balances, centrifuges, incubators, fume hoods and anything calibrated need checking, and often recalibrating, in the new location. A fume hood in particular is only proven in the room it now sits in, so containment testing is part of the move.

**What is the first step in moving laboratory equipment?**

Inventory and triage. Dispose of unused equipment and unidentifiable chemicals before the move rather than paying to relocate them, and start decontamination early because it sits on the critical path.

**Do instruments need a decontamination certificate?**

Yes. Movers and service engineers are entitled to refuse equipment without a signed statement of what it was exposed to and how it was treated, and most will.

**How should frozen samples be moved?**

As a separate plan made first, using dry ice, validated shippers or powered transport with somebody accompanying them and a written contingency. Freezer contents are usually the least replaceable thing in the building.

**Does equipment need recalibrating after a move?**

Balances, centrifuges, incubators and anything calibrated should be checked and usually recalibrated in the new location, and fume hoods must be containment tested in the room they now occupy.

## LIMS against ELN

The two categories are built on different data models, and that is the whole distinction. A laboratory information system is organized around samples moving through defined tests; an electronic lab notebook is organized around experiments and the reasoning in them. Buy the one whose model matches your work and the system feels obvious; buy the other and you will spend years forcing your work into a shape it does not have. This page sets out the difference and when both are justified.

1. **The object at the center**: In a laboratory information system the central object is a sample with tests attached; in a notebook it is an experiment with observations attached. Ask which of those sentences describes what your laboratory does all day, and you have your answer in most cases.
2. **Repeatability against narrative**: A system of record excels where the same method runs many times and results must be comparable, reviewed and released. A notebook excels where each experiment differs and the reasoning matters as much as the numbers. Neither does the other's job well however many modules are added.
3. **Who the output is for**: A laboratory information system produces certificates, trends and released results for a customer or a regulator. A notebook produces a defensible record of what was done and why, for colleagues, patent attorneys and future you. Different audiences, different shapes.
4. **Structure against freedom**: A system of record enforces structure, which is its value and its friction. A notebook permits freedom, which is its value and its weakness, since freely written records are harder to search and to aggregate. Templates in a notebook recover some structure without the rigidity.
5. **When you need both**: Organizations that both develop methods and run them routinely genuinely need both, with an explicit boundary and a shared sample identity across it. That boundary has to be decided rather than discovered, or the two systems will hold contradictory versions of the same sample.

**How does the LIMS data model differ from an ELN's?**

A LIMS is organized around samples moving through defined tests and produces released results; an ELN is organized around experiments and the reasoning in them. The data model is the difference, and modules do not change it.

**Which one should a testing laboratory buy?**

A LIMS. The work is the same methods run many times on many samples, with results that must be comparable, reviewed and released, which is exactly what that model is built for.

**Do I need both a LIMS and ELN?**

If you both develop methods and run them routinely, usually yes, with an explicitly decided boundary and a shared sample identity across it. Discovering the boundary later means two systems disagreeing about one sample.

**Does an ELN need validating?**

Where it holds data supporting a regulated submission or product, yes. The obligation follows the data, not the category of product holding it.

## LIMS cost

Almost every laboratory information system budget is wrong in the same direction, because it is built from the license quotation and the license is the smallest durable line. A budget that holds is built over five years and includes the internal effort, which is the cost nobody invoices and the one that most often derails the project. This page sets out the line items in the order they arrive.

1. **License, and how it scales**: Per named user or per concurrent session, annual subscription or perpetual with maintenance, plus whatever modules sit outside the base. Model it at the headcount you expect in year five rather than year one, because growth in this line is where subscriptions surprise people.
2. **Implementation, the largest first-year number**: Configuration, workflow build, report templates, data migration and training are quoted as a project and routinely exceed the first year of license. Insist on a scope this is priced against, since an implementation quoted against a vague scope will be re-quoted later.
3. **Interfaces, priced per instrument**: Each instrument interface is its own small project, priced individually and dependent on whether a supported driver exists for that make, model and firmware. The instrument list is the real driver of this line and is often produced too late.
4. **Validation and requalification**: In a regulated environment, qualification documents have to be produced and then maintained through every upgrade. Ask whether the vendor sells a validation pack and what each upgrade costs in requalification, because that recurs for the life of the system.
5. **Internal effort, the line nobody invoices**: Your own people will spend substantial time on requirements, configuration decisions, testing, data cleansing and training. It does not appear on any quotation and it is real, so put a figure on it and staff it deliberately rather than assuming it will absorb into existing jobs.

**What drives LIMS cost most, and is free LIMS software ever cheaper?**

Instrument count and workflow complexity, then regulatory scope. User count drives the license line but not the implementation, which is usually the larger first-year number.

**Is the licence the main part of the LIMS software price?**

Rarely in year one. Implementation, interfaces, validation and internal effort typically exceed it, and internal effort is the line nobody invoices and most budgets omit.

**Cloud or on-premises: which lowers the cost of LIMS?**

Over five years the totals are usually closer than either vendor suggests. The real differences are control, data location and who carries the upgrade and infrastructure burden.

**What should I ask about upgrades?**

What an upgrade costs in requalification in a validated environment, whether configuration survives it, and how often upgrades are mandatory. This recurs for the life of the system.

## Flow cytometry services

Outsourcing flow cytometry works well when the provider designs or validates the panel, runs standardized instruments and returns both the analysis and the underlying files. It works badly when samples arrive degraded, when the controls were not agreed in advance, or when only a summary comes back. The logistics matter as much as the cytometry. This page covers what to agree before the first sample ships.

1. **Panel design and who owns it**: Establish whether the provider designs the panel, validates one you supply, or runs an existing catalog panel, and who owns the design afterwards. A panel is a real piece of work and its ownership should not be discovered at the end of a study.
2. **Sample type, stability and shipping**: Fresh whole blood, isolated cells and fixed samples each have a stability window and a shipping requirement, and a sample outside its window produces data that looks real and is not. Agree the window, the shipper, the courier and what happens to a late arrival before anything ships.
3. **Controls, agreed and paid for**: Single stain compensation controls, fluorescence minus one controls where gates are ambiguous, viability discrimination and an isotype or biological negative where appropriate. These consume samples and instrument time and must be in the scope rather than assumed.
4. **Instrument standardisation across a long study**: For a study running months, ask how the provider standardises instruments over time with beads and how a service event or a laser replacement is handled. Without that, drift becomes an apparent biological effect.
5. **Data delivery, including the raw files**: Require the raw acquisition files as well as the analysis, with the gating strategy documented. A summary table alone cannot be re-examined, and re-analysis is exactly what you will want when a result is surprising.

**What should a flow cytometry service deliver?**

The analysis, the documented gating strategy and the raw acquisition files, along with the control data. A summary table alone cannot be re-examined when a result is surprising.

**What controls should be in the scope?**

Single stain compensation controls, viability discrimination, fluorescence minus one controls where gates are ambiguous, and an appropriate negative. They consume sample and instrument time, so they must be priced rather than assumed.

**How do samples need to be shipped?**

According to the stability window for that sample type, with an agreed shipper and courier and a written plan for a late arrival. A sample analyzed outside its window produces plausible data that is not valid.

**How is drift handled over a long study?**

By standardising the instrument with beads on a defined schedule and recording it, with a documented approach to service events. Without that, instrument drift appears as a biological effect.

**Who owns flow cytometry panel design, and the flow cytometry panels themselves?**

Whoever the agreement says, so settle it before the first sample ships. Establish whether the provider designs the panel, validates one you supply or runs an existing catalog panel, and who owns the design afterwards: a panel is a real piece of work, and its ownership should not be discovered at the end of a study.

## RNA-seq analysis

Most disappointing transcriptome experiments were decided before any sequencing happened, in the choice of replicates, library type and how samples were distributed across batches. The analysis cannot rescue a design that confounded the comparison, and no depth substitutes for replication. This page covers the decisions that matter, in the order they are made, and which parts of the work are worth handing to somebody else.

1. **Replicates beat depth**: Statistical power for differential expression comes from biological replication far more than from reads per sample, and technical replicates measure the machine rather than the biology. When budget is fixed, more biological replicates at moderate depth almost always beats fewer at high depth.
2. **Library type follows the question**: Poly-A selection suits intact eukaryotic messenger RNA; ribosomal depletion suits degraded, bacterial or non-polyadenylated transcripts. Stranded libraries resolve overlapping features and antisense transcription and are now the sensible default. These choices are made before sequencing and cannot be changed after.
3. **Randomise across batches**: If every treated sample is prepared on one day and every control on another, treatment and batch are confounded and no analysis can separate them. Distribute groups across preparation batches and sequencing runs, and record which sample went where.
4. **Reference and annotation are a choice**: The genome build and annotation release determine what can be quantified, and mixing releases across a project produces results that cannot be compared. Fix both at the start, record them, and use them throughout.
5. **Normalisation and the statistical method**: Counts must be normalised for depth and composition before comparison, and the established differential expression packages handle that along with dispersion estimation. Spreadsheet fold-changes on raw counts are not an analysis, whatever the numbers look like.

**How many replicates do I need for RNA seq?**

As many biological replicates as the budget allows, because power for differential expression comes from replication far more than from sequencing depth. Technical replicates measure the instrument rather than the biology.

**Poly-A selection or ribosomal depletion?**

Poly-A selection for intact eukaryotic messenger RNA; ribosomal depletion for degraded material, bacterial RNA or non-polyadenylated transcripts. The choice is made before sequencing and cannot be revisited.

**What should an outsourced analysis deliver?**

The count matrix, the software versions and parameters, the quality control report and the raw data, alongside the differential expression results. Without those the analysis cannot be reproduced.

**How do I avoid batch effects?**

Distribute treatment groups across preparation batches and sequencing runs rather than processing each group together, and record which sample went where. A confounded design cannot be fixed analytically.

## Ion exchange chromatography

Anion exchange separates on charge, and charge depends on pH, which makes the buffer pH the single decision that determines whether the method works. Choose it correctly and the target binds while much of the contamination flows through; choose it wrongly and either nothing binds or everything does. This page covers picking the pH, the resin and the elution, and why this step is usually where purity is won.

1. **Set pH from the isoelectric point**: A protein carries net negative charge above its isoelectric point and will bind an anion exchanger there. Working roughly a unit above the target's isoelectric point is the usual starting point, adjusted for what the contaminants do at the same pH, since the separation is about the difference rather than the target alone.
2. **Strong or weak exchanger**: A strong exchanger keeps its charge across the working pH range and is the predictable default. A weak exchanger changes charge with pH, which offers extra selectivity and extra complexity. Start strong and move to weak only if the separation genuinely needs it.
3. **Ionic strength on load**: A sample loaded in too much salt will not bind however correct the pH, and this is the commonest practical failure. Dilute, desalt or exchange the buffer before loading, and check conductivity rather than assuming it from the recipe.
4. **Elute by salt or by pH**: A salt gradient is the standard and is easy to control and to scale. A pH gradient can give sharper selectivity and is harder to run reproducibly. Shallow gradients resolve better and take longer, and a step elution is fastest where the separation is easy.
5. **Scout before you commit**: A short linear gradient run tells you where the target and the contaminants elute, and that single experiment usually sets the whole method. Skipping it means guessing at conditions that a half-day run would have established.

**What pH should I use for anion exchange?**

Start roughly a unit above the target protein's isoelectric point so it carries net negative charge and binds, then adjust based on what the contaminants do at that pH. The separation depends on the difference, not the target alone.

**Why is my protein not binding?**

Most often the load contains too much salt, so check conductivity rather than trusting the recipe. The other common cause is a pH below the protein's isoelectric point, where it is not negatively charged.

**Strong or weak anion exchanger?**

Strong exchangers hold their charge across the working pH range and behave predictably, which makes them the sensible default. Weak exchangers add selectivity and complexity and are worth trying only when the separation needs it.

**Salt gradient or pH gradient?**

Salt gradients are standard, easy to control and scale well. pH gradients can be more selective and are harder to run reproducibly. Run a short scouting gradient before fixing either.

**Is histidine tag affinity chromatography a purification?**

It is a first step. A family of host proteins co-purifies reliably on metal affinity, so a second orthogonal step is what produces usable material. Choose the tag with its removal in mind.

## Equipment calibration

Calibration is usually run as a series of annual appointments, and that is why it costs more than it should and catches less than it should. A program decides what genuinely needs calibrating, sets intervals from use and consequence rather than the calendar, and puts cheap internal checks between expensive external visits. This page sets out how to build one.

1. **Decide what is calibrated and what is verified**: Instruments whose reading forms part of a result need calibration; instruments used qualitatively need a functional check. Calibrating everything wastes money, and calibrating nothing means results nobody can defend. Write the list and the reason against each entry.
2. **Set intervals from use and consequence**: An instrument used constantly for a result somebody relies on earns a shorter interval than one used occasionally as an indicator. A uniform annual interval across a laboratory is a purchasing convenience rather than a technical decision, and it is both wasteful and risky at the two ends.
3. **Put internal checks between external visits**: A daily check weight on a balance, a reference thermometer in an incubator, a quick gravimetric check on a pipette: these cost minutes and catch drift long before the annual service would. They are what make a long external interval defensible.
4. **Require certificates that report numbers**: As-found and as-left values, the acceptance limits applied and where they came from, the ambient conditions and the reference used. A certificate reporting only a pass has withheld the information that lets you see an instrument drifting.
5. **Decide where accreditation is required**: Accredited calibration carries third party assessment of the provider's competence and traceability, and regulated or accredited laboratories generally need it. A research laboratory often does not, and can buy competent unaccredited service far more cheaply where the certificate still reports real numbers.

**What laboratory equipment needs calibration?**

Anything whose reading forms part of a result. Instruments used qualitatively need a functional check instead. Write the list with a reason against each entry rather than calibrating everything or nothing.

**How often should equipment be calibrated?**

At an interval set by how heavily it is used and what depends on the result, with cheap internal checks between external visits. A uniform annual interval is a purchasing convenience, not a technical decision.

**What should a calibration certificate show?**

As-found and as-left values, the acceptance limits and their source, the ambient conditions and the reference used. A pass or fail statement alone hides the drift you are paying to detect.

**Do I need accredited calibration?**

Regulated and accredited laboratories generally do. Research laboratories often do not and can buy competent unaccredited service more cheaply, provided the certificate still reports measured numbers.

## TOC analysers

A total organic carbon analyser oxidises the carbon in a sample and measures what comes off, and the oxidation method decides which samples it can handle. That single choice determines whether the instrument suits clean pharmaceutical water or difficult environmental and industrial samples, and it is not adjustable afterwards. This page covers choosing the method, the range and the deployment.

1. **Oxidation method decides the matrix**: High temperature combustion oxidises almost anything including particulates and difficult compounds and suits dirty, high-carbon and saline samples. Ultraviolet persulfate oxidation is gentler, more sensitive at low concentrations and suits clean water, and it struggles with particulates and some refractory compounds.
2. **Range and where your samples sit**: Instruments are optimized for a range, and one covering a wide span is usually a compromise at both ends. Establish the concentrations you actually measure and buy for that band rather than for the widest specification available.
3. **How inorganic carbon is handled**: Total carbon includes carbonate, so the instrument must either remove inorganic carbon by acidification and sparging or measure and subtract it. For samples with high carbonate this handling is a major source of error and deserves a direct question.
4. **Online or laboratory instrument**: Online analysers monitor a water system continuously and give immediate warning of an excursion; laboratory instruments handle varied samples and give a documented result per sample. Water system monitoring usually justifies online, and varied sample types usually do not.
5. **Calibration, standards and suitability**: Calibration standards, and for pharmaceutical water a system suitability check with a readily and a poorly oxidised compound, are what demonstrate the instrument is fit. Confirm the instrument supports the suitability procedure your application requires.

**Which oxidation method should total organic carbon analyzers use?**

High temperature combustion for dirty, saline or particulate-containing samples; ultraviolet persulfate for clean water at low concentrations. The method determines which matrices the instrument can handle and cannot be changed later.

**Online total organic carbon analyzer or a laboratory one?**

Online for continuous monitoring of a water system with immediate excursion warning; laboratory instruments for varied sample types and documented per-sample results. Many pharmaceutical sites run both.

**How does a total carbon analyzer handle inorganic carbon?**

Either removed by acidification and sparging before measurement, or measured and subtracted. With high carbonate samples this handling is a major error source and worth asking about directly.

**Can TOC be used for pharmaceutical cleaning validation?**

It is widely used as a non-specific residue indicator, valued for speed and breadth. Because it reports carbon without identifying the source, a recovery study relating it to the actual residue is required.

**Why does my pH reading drift?**

Almost always the electrode. The glass response slows, the junction clogs and the reference electrolyte depletes. Record the calibration slope each time: a slope drifting from ideal is the electrode telling you it is ageing.

**Are organic carbon results comparable between instruments?**

Not directly. Combustion and reagent-based oxidation recover different fractions, so the method belongs beside the number. For pharmaceutical water the method and its suitability test are specified.

## DNA ladders

A DNA ladder is a mixture of fragments of known length run beside your samples so that a band can be assigned a size, and choosing one is a matter of matching its range and spacing to the gel you actually run. A good number of laboratories use one ladder for everything and read sizes from a region where it has no resolution. This page covers choosing well and using the ladder for more than size.

1. **Match the range to the fragments**: A ladder spanning a wide range has coarse spacing in the middle; one designed for small fragments resolves them finely and says nothing above its top band. Choose from the sizes you actually assign, and keep a second ladder for the other end rather than extrapolating.
2. **Spacing matters more than the number of bands**: Closely spaced bands around your fragment size let you assign a length with confidence; a ladder whose nearest bands are far apart gives you a range, not a size. Look at the spacing near your region of interest rather than the band count.
3. **Reference bands give you quantity**: Many ladders load a known mass into designated bands, so comparing your band's intensity to a reference band estimates quantity as well as size. It is approximate and it is free, and it is the quickest sanity check on a preparation's concentration.
4. **Ready-to-load or concentrate**: Ready-to-load formats include loading dye and are quicker and more consistent; concentrated stocks are cheaper per lane and need dilution and dye. For routine work the ready format usually pays for itself in avoided pipetting errors.
5. **Store it properly and replace it**: Repeated freeze and thaw cycles and prolonged warmth degrade a ladder, and a smeared or partial ladder makes every gel it appears on uninterpretable. Aliquot on arrival, and discard a ladder that no longer runs cleanly rather than working around it.

**How do I choose a DNA ladder?**

Match its range and, more importantly, its band spacing to the sizes you actually assign. A wide-range ladder has coarse spacing in the middle, which turns a size assignment into a guess.

**Can I use a DNA ladder to estimate concentration?**

Approximately, where the ladder loads a known mass into designated bands. Comparing your band's intensity to a reference band is a quick and free sanity check, though not a substitute for proper quantitation.

**Why does my plasmid not run at its expected size?**

Conformation changes migration, so supercoiled, relaxed and linear forms of the same plasmid run differently. Linearise before sizing, and treat gel sizing as an estimate.

**Why has my ladder stopped giving clean bands?**

Usually degradation from repeated freeze and thaw cycles or prolonged warmth. Aliquot on arrival and discard a ladder that no longer runs cleanly rather than interpreting around it.

**Which ladder range should I buy?**

One whose bands bracket your fragment, since resolution is best mid-range. A wide-range set is convenient and resolves poorly at both ends; choose the ladder and the gel percentage together.

**Can I use a double-stranded ladder on a denaturing gel?**

No. Its strands separate and run at a different apparent size, so sizes read from it are wrong. Use a ladder supplied as single strands of known length.

## LNP formulation services

Formulating messenger RNA into lipid nanoparticles is a process question, not a mixing question: the same lipids and the same RNA produce different particles depending on how they are combined, and the resulting particle is what determines whether anything is delivered. Providers differ in the process they run and in what they measure afterwards. This page covers what to specify and what characterisation to insist on.

1. **Lipid composition and its license position**: The ionisable lipid dominates delivery behavior and is frequently proprietary and licensed. Establish which lipid is used, whether the composition is the provider's or yours, and what the license terms are for your intended use, because that position travels with the program.
2. **The mixing process is the product**: Particle size and uniformity come from how the streams are combined, and microfluidic and impingement approaches give different particles from identical inputs. Ask which process is used and whether the same process is available at the scale you may eventually need.
3. **Encapsulation efficiency and payload**: Unencapsulated RNA is degraded quickly and delivers nothing, so encapsulation efficiency and RNA content per particle should be measured and reported per batch rather than quoted as a process capability.
4. **Size, dispersity and charge**: Particle size distribution and dispersity affect biodistribution and reproducibility, and apparent surface charge affects tolerability. These should be reported per batch with the method, since they are the properties most likely to shift between runs.
5. **Stability and storage**: Establish the storage condition and the demonstrated stability period, including whether a cryoprotectant is used and how freeze and thaw affects the particle. A formulation that has not been stability tested at your storage condition is an unknown.

**What should an LNP formulation service report per batch?**

Encapsulation efficiency, RNA content, particle size distribution and dispersity, apparent surface charge, and the storage condition with demonstrated stability, each with the method used.

**Why does the mixing process matter?**

Because identical lipids and RNA give different particles depending on how the streams are combined. Ask which process is used and whether it is available at the scale you may eventually need.

**Does the ionisable lipid carry licence obligations?**

Frequently. The ionisable lipid dominates delivery behavior and is often proprietary, with terms that attach to your program. Establish the position before committing to a composition.

**Can research formulations support a regulated application?**

No. Raw material grades, process and testing all differ and cannot be applied retrospectively. Ask what a compliant route requires early, since compliant lipid is a long-lead item.

## Cell and gene therapy CDMOs

Cell and gene therapy manufacture differs from conventional biologics in ways that change how a provider is chosen: batches are small and often patient-specific, the starting material may be a person's cells with a fixed schedule, and analytical methods are less standardized. Capacity and experience in your exact modality matter more than price. This page covers the questions that separate providers.

1. **Experience in your modality specifically**: Viral vector production, autologous cell processing, allogeneic manufacture and gene-edited products are different operations with different equipment and different problems. Ask how many programs in your exact modality the site has run, and ask to speak to one of those sponsors.
2. **Suite capacity and what a slot means**: Capacity is measured in suites and in the changeover between campaigns, and for autologous products in slots per period. Establish what is contractually reserved, what happens if your timeline slips, and what happens if theirs does.
3. **Starting material logistics**: Where the starting material is a patient's cells, the chain of identity and the chain of custody from collection to manufacture and back are part of the service. Ask how these are managed, what the systems are, and what happens when a shipment is delayed.
4. **Analytics and potency**: Potency assays for these products are frequently the hardest part of the program and are less standardized than for conventional biologics. Establish who develops them, who owns them, and how they transfer, because this is where timelines slip most often.
5. **Regulatory experience and inspection history**: Ask which authorities have inspected the site, when, for which modality, and how the site supported previous sponsors' filings. Experience of the regulatory path in your modality is worth more than nominal capacity.

**How do I choose a cell and gene therapy CDMO?**

On experience in your exact modality, on real suite capacity and what a reserved slot contractually means, on starting material logistics, and on analytical capability. Price is rarely the deciding factor in this field.

**Why are analytics such a problem here?**

Potency assays for cell and gene therapy products are less standardized than for conventional biologics and are frequently the hardest part of a program. Establish who develops them, who owns them and how they transfer.

**What makes technology transfer risky in cell and gene therapy manufacturing?**

Processes are often less defined and more of the knowledge sits with people rather than in documents, so transfer commonly reveals gaps. A provider that plans for that is a better partner than one that assumes a clean handover.

**What does a manufacturing slot actually guarantee?**

Only what the contract says. Establish what is reserved, what happens if your timeline slips and what happens if the provider's does, because capacity in this field is genuinely constrained.

## Genome sequencing services

Genome sequencing is priced per sample and the price conceals three choices that decide whether the data answers your question: the read technology, the depth, and how much analysis is included. Short and long reads see different things, and a project that needs structural variants will not find them in short reads however deep the coverage. This page covers the choices in the order they constrain each other.

1. **Short or long reads**: Short reads are cheap, accurate per base and excellent for small variants. Long reads resolve repeats, structural variation and phasing and cost more per base. A project that could involve structural change or repetitive regions needs long reads or a combination, and no amount of short-read depth substitutes.
2. **Depth follows the variant class**: Germline variant calling needs far less depth than detecting a variant present in a small fraction of cells, and de novo assembly is different again. State the variant class and the lowest fraction you must detect, and let that set the depth rather than accepting a default.
3. **Input quality decides the library**: Long-read library preparation needs high molecular weight DNA that ordinary column extraction damages. If long reads are the plan, the extraction method is part of the plan, and a sample already extracted on a column may not be usable.
4. **Analysis scope, stated explicitly**: Reads only, aligned data, variant calls, annotation or interpretation are different products at different prices. Establish which you are buying, in which file formats, and against which reference build and annotation release.
5. **Data volume, delivery and retention**: Genome data is large. Agree how it is delivered, how long the provider keeps it, what a second copy costs, and where it is stored, because the storage question is a real cost and, for human data, a compliance question.

**Short reads or long reads?**

Short reads for small variants at low cost; long reads for structural variation, repeats and phasing. A question that could involve structural change needs long reads, and short-read depth does not substitute.

**What depth do I need?**

It depends on the variant class and the lowest fraction you must detect. Germline calling needs far less than low-fraction detection, and de novo assembly is different again. State the requirement and let it set the depth.

**Does extraction method matter?**

For long reads, considerably. Library preparation needs high molecular weight DNA that ordinary column extraction damages, so the extraction method has to be planned with the sequencing.

**What analysis is included?**

It varies from raw reads to full interpretation. Establish the scope, the file formats and the reference build and annotation release explicitly, because this is where similar-looking quotes differ most.

**Is a whole genome sequencing kit an alternative to a service?**

Only where the instrument and the analysis already exist in house. A library kit makes the library; it does not make the run, the depth or the interpretation, and the cost of a sequencer idling between projects is what sends most laboratories to a service. Where there is a run to fill every week, the kit is cheaper per sample.

**When does rapid whole genome sequencing change the answer?**

When a decision waits on it. Turnaround is bought with dedicated instrument time, a smaller batch and an analysis pipeline that runs unattended, and all three cost more per sample. For a research cohort it buys nothing; where a result informs care or an outbreak response, the speed is the point of the assay.

## NGS testing services

Next generation sequencing testing covers everything from a small targeted panel to a whole genome, and providers differ far more in validation than in chemistry. What decides whether a result is usable is the panel design, the validated limit of detection, and whether the laboratory operates under an accreditation appropriate to the use. This page covers choosing a provider and reading what comes back.

1. **Panel design against your genes of interest**: A targeted panel sequences a defined gene set very deeply and cheaply, and it sees nothing outside that set. Check the panel's content against your genes and, importantly, against the specific regions within them, since coverage within a gene is rarely uniform.
2. **Validation and the limit of detection**: Ask what the laboratory validated, on what sample types, and at what lowest detectable variant fraction. A detection claim without a stated validation is a capability statement rather than a specification you can rely on.
3. **Accreditation appropriate to the use**: Research testing and testing that informs a clinical decision sit under different regimes, and in the United States clinical testing requires a laboratory operating under the clinical laboratory improvement regime. A research result cannot be repurposed for a clinical decision because it was technically good.
4. **Sample type and input requirement**: Fresh tissue, fixed tissue, blood and cell-free material each have different input requirements and different achievable quality. Declare the sample type at quoting, since fixed material in particular constrains what is possible.
5. **Turnaround, repeat policy and failures**: Establish the turnaround, what happens when a sample fails quality control, whether a repeat is included and whether you are charged. Sample failures are routine in this field and the policy matters.

**What should I check before choosing an NGS testing provider?**

The panel content against your genes and the regions within them, the validated limit of detection and the sample types it was validated on, the accreditation appropriate to your use, and the policy on sample failures.

**What is the difference between research and clinical NGS testing?**

They operate under different regimes. In the United States, testing that informs a clinical decision must come from a laboratory under the clinical laboratory improvement regime; a research result cannot be repurposed however good it is technically.

**What should an NGS report contain?**

The regions covered and those that failed coverage, variants with their fractions, the limit of detection applied, and an explicit statement of what the test does not cover. Absence of a finding means nothing without that.

**Can fixed tissue be used?**

Usually, with constraints. Fixation fragments and chemically damages nucleic acid, so yields are lower and some approaches are unavailable. Declare the sample type at quoting rather than at the laboratory.

## LC-MS and mass spectrometry services

Liquid chromatography coupled to mass spectrometry is two instruments and the coupling between them, and the analyser decides what questions can be asked. Targeted quantitation of known compounds and discovery of unknown ones need different instruments, and buying the wrong one produces an expensive machine that does the other job badly. This page covers choosing an analyser and, for most laboratories, scoping the work with a provider instead.

1. **Targeted quantitation wants a triple quadrupole**: For measuring known compounds at low concentrations in complex matrices, a triple quadrupole monitoring specific transitions gives the sensitivity and the dynamic range, and it is the workhorse of quantitative bioanalysis and residue testing.
2. **Discovery wants high resolution**: Identifying unknowns needs accurate mass and fragmentation, which means a time-of-flight or orbital trap analyser. These also quantify, less well at the very bottom of the range, and they generate far more data that somebody has to handle.
3. **Ionisation follows the analyte**: Electrospray suits polar and ionisable compounds; atmospheric pressure chemical ionisation suits less polar ones. Some analytes ionise poorly in both and need derivatisation. This is decided by chemistry rather than by instrument brand.
4. **Matrix effects are the real difficulty**: Co-eluting matrix components suppress or enhance ionisation, which moves results without any obvious sign. Stable isotope labeled internal standards are the proper defense, and a method without one in a complex matrix should be treated with caution.
5. **Scoping an outsourced method**: Give the provider the analyte, the matrix, the concentration range and the required limit, and ask what internal standard is used, what validation has been performed in that matrix, and what the certificate reports. A quote without those is not comparable with another.

**Which mass analyser do I need?**

A triple quadrupole for targeted quantitation of known compounds at low levels; a high resolution analyser for identifying unknowns. They are different instruments for different questions and each does the other's job poorly.

**What are matrix effects?**

Suppression or enhancement of ionisation by co-eluting sample components, which shifts results without any obvious sign. Stable isotope labeled internal standards are the proper defense in complex matrices.

**Should I buy an LC-MS or use a mass spectrometry analysis service?**

Outsource occasional work: the instrument is the smaller cost beside service, consumables and an experienced operator. Ownership makes sense when the workload is continuous and the method is yours.

**What should an LC-MS quote specify?**

The analyte, matrix, concentration range and required limit, the internal standard used, the validation performed in that matrix, and what the certificate reports. Without those, two quotes are not comparable.

**Which technique does my sample need?**

Gas phase separation if the analyte is volatile and thermally stable or can be derivatised; liquid phase for everything polar, large or fragile. The analyte decides, and many laboratories own both.

**Is method development included when I buy an instrument?**

Rarely, and it is most of the work. If a validated method exists for your analyte and matrix, buy the instrument that runs it; if not, put the months of development into the business case.

## Cell banking services

A cell bank exists so that every future batch starts from the same material, and its value is entirely in the testing and the records that accompany it. A freezer full of vials with no characterisation is not a bank, it is an inventory. This page covers the two-tier structure, the testing each tier owes and the storage decisions that decide whether the bank survives a bad night.

1. **Two tiers, and why**: A master bank is laid down once and drawn on rarely; a working bank is made from one master vial and is what routine production consumes. The structure exists so the master is protected and so every working bank has a traceable common origin, and it is worth adopting even in research settings.
2. **Testing the master properly**: Identity confirming the line is what it claims, sterility, mycoplasma, and adventitious agent testing to a panel appropriate to the origin and the intended use. This is the expensive part and it is the part that makes the bank worth having.
3. **Characterisation beyond safety**: Viability and recovery on thaw, growth characteristics, and for a production line the expression and quality attributes. A bank that is clean but does not perform is still a problem, and finding that out at the first working bank is expensive.
4. **Split the storage**: Hold the bank in at least two physically separate locations with independent power and monitoring. Freezer failures and building incidents happen, and a bank in one place is a single point of failure for everything that depends on it.
5. **Records that outlive people**: Passage history, the medium and reagents used, the date, the operator, the vial map and the test results. A bank whose provenance cannot be reconstructed cannot support a regulatory filing and is hard to trust even in research.

**Why have a master cell bank and a working cell bank?**

So that the master is protected and drawn on rarely, while routine work consumes working banks made from single master vials. Every batch then traces to one characterised origin.

**What testing does a cell bank need?**

Identity, sterility, mycoplasma and adventitious agent testing appropriate to the origin and intended use, plus viability, recovery and performance characterisation. The testing is what makes it a bank rather than an inventory.

**Where should a cell bank be stored?**

In at least two physically separate locations with independent power and monitoring. A single-site bank is a single point of failure for every program depending on it.

**Do research laboratories need cell banks?**

They benefit greatly. Authenticating the line, testing for mycoplasma and laying down a small two-tier bank costs little against the years lost to a contaminated or misidentified line.

## Antibody characterisation

Characterisation is easy to buy and easy to over-buy, because a provider will happily run a long panel on candidates that a short one would have eliminated. The efficient order runs cheap, discriminating measurements first and expensive, informative ones on the survivors. This page sets out that order and what each measurement is actually good for.

1. **Specificity before affinity**: An antibody that binds tightly to the wrong thing is worse than one that binds moderately to the right thing. Screen against a counter-panel of close relatives and the species you will work in before spending on kinetics, since this eliminates candidates fastest and most cheaply.
2. **Affinity and kinetics with the method stated**: Surface plasmon resonance and biolayer interferometry give association and dissociation rates as well as the equilibrium constant, and the off rate is frequently what matters functionally. Ask which method, which orientation and which regeneration conditions, because all three move the numbers.
3. **Epitope binning when you have several**: Grouping candidates by whether they compete tells you how many distinct epitopes you actually have and which pairs could work together in a sandwich assay or a bispecific. It is worth doing once the panel is down to a manageable number.
4. **Function in the assay that matters**: Binding does not predict blocking, agonism, internalisation or staining behavior. A functional assay in the intended format is the measurement that decides the candidate, and it should be planned even if it is run in house.
5. **Developability, before you commit**: Thermal stability, aggregation propensity, expression level and behavior at concentration decide whether a candidate can be made and stored. These are cheap to screen and expensive to discover late.

**What should I measure first on an antibody panel?**

Specificity against close relatives and the working species. It eliminates candidates fastest and most cheaply, and a tight binder to the wrong target is worse than a moderate binder to the right one.

**Are affinity values comparable between providers?**

Only when the method, instrument, orientation, buffer, temperature and analysis model are the same. Ask for all of those alongside the number, and for the raw traces.

**What is epitope binning for?**

Grouping candidates by competition, which tells you how many distinct epitopes you have and which pairs can work together in a sandwich assay or a bispecific format.

**Why screen developability early?**

Because thermal stability, aggregation propensity and expression level decide whether a candidate can be made and stored at all. They are cheap to screen and very expensive to discover after lead selection.

## Biologics manufacturing

Biologic manufacture divides into three stages that fail in different ways and are costed very differently, and organizations buying it for the first time consistently under-weight two of them. Upstream gets the attention; downstream consumes the yield; fill finish is where a batch is most often lost at its most valuable. This page sets out where cost and risk actually sit and what to settle before choosing a site.

1. **Upstream: the cell culture**: Cell line, medium, feed and the parameters held during production set titre and product quality. It is the stage most amenable to optimisation and the one where a contamination event costs the whole batch, which is why single use systems have become the default at moderate scale.
2. **Downstream: capture through polish**: Chromatography, filtration, viral clearance and formulation. Every step costs yield, and resin and membrane consumption is a major recurring cost. Downstream is usually where a process is capacity constrained even though upstream gets the attention.
3. **Fill finish: small, late and high risk**: Filling into vials or syringes happens at the end, on the most valuable material in the process, and it has its own sterility assurance requirements and its own specialist sites. Capacity here is genuinely constrained and lead times are long, so it is booked far earlier than most organizations expect.
4. **Single use against stainless**: Single use removes cleaning and sterilisation validation and shifts cost into consumables and a supply chain you do not control. Stainless is cheaper per run at high utilization and carries validation and utilities overhead. Utilization decides it, not preference.
5. **Comparability across changes**: Any change of site, scale or process needs comparability demonstrated between what came before and after, and that exercise is a project in itself. Plan changes to happen at points where comparability is cheapest rather than when convenient.

**What are the stages of biologics manufacturing?**

Upstream cell culture, downstream purification and formulation, and fill finish into the final container. They fail in different ways and are costed very differently, and the last two are routinely under-weighted.

**Single use or stainless steel?**

Single use where utilization is moderate and cleaning validation is a burden; stainless where utilization is high and the process is stable. Utilization decides it rather than the capital cost.

**Why is fill and finish biologics work a bottleneck?**

Because capacity is genuinely constrained, lead times are long, and it happens at the end on the most valuable material with its own sterility assurance requirements. Book it far earlier than feels necessary.

**What is comparability?**

The demonstration that product made after a change of site, scale or process is equivalent to what came before. It is a project in itself, so plan changes for points where it is cheapest to show.

## 3D cell culture

Three dimensional culture is adopted because cells in it behave more like cells in tissue, and it is abandoned because the readouts are harder. Reagent penetration, imaging depth and lysis efficiency all change when the sample is a solid object rather than a monolayer, and a laboratory that chooses a format without solving the readout first ends up with better biology it cannot measure. This page covers both halves.

1. **Spheroids for scale and consistency**: Self-assembled aggregates in low attachment plates or hanging drops are the simplest route, reproducible in size when seeded carefully and compatible with plate-based screening. They are the sensible starting point for most laboratories moving off flat plastic.
2. **Organoids for architecture**: Stem cell derived structures that self-organize into tissue-like architecture answer questions spheroids cannot, at the cost of longer culture, higher variability and considerably more expensive media. They are a research program rather than a plate format.
3. **Hydrogels and scaffolds for the matrix question**: Where the matrix itself matters, a defined synthetic hydrogel gives control and reproducibility, while animal-derived extracts are closer to native composition and vary substantially between lots. Choose from whether you need control or physiological relevance, and test lots either way.
4. **Solve the readout before the first experiment**: Viability reagents penetrate a solid structure slowly and incompletely, lysis is harder, and standard imaging cannot see into the middle. Decide whether the readout is imaging, a penetration-validated assay, or dissociation followed by a conventional measurement, and validate it on your own structures.
5. **Control the size**: Larger structures develop gradients of nutrients and oxygen and eventually necrotic cores, which is either the phenomenon you are studying or a confound. Either way, size has to be controlled and measured rather than left to the seeding density.

**Which 3D cell culture format should I use?**

Spheroids for scale, consistency and plate-based work; organoids where tissue-like architecture is the point, accepting longer culture and higher variability; hydrogels or scaffolds where the matrix itself is part of the question.

**Why are 3D assays harder to read out?**

Reagents penetrate solid structures slowly and incompletely, lysis is harder, and standard imaging cannot see into the middle. The readout must be chosen and validated before the first experiment rather than after.

**How do I handle matrix lot variation?**

Test each lot against a retained reference before committing a study and record which lot produced which data. Animal-derived extracts vary in composition and stiffness in ways that look like biology.

**Does structure size matter?**

Considerably. Larger structures develop nutrient and oxygen gradients and necrotic cores, which is either your subject or a confound. Control and measure size rather than leaving it to seeding density.

## Ultra-low freezers

An ultra-low freezer is a purchase where the box is the least important part. What decides whether you lose samples is where it is sited, how the room handles the heat it rejects, whether the alarm reaches a person who can act, and whether anyone has decided in advance what happens when it fails. Freezers fail, and the ones that cost a laboratory years are the ones nobody planned for. This page is mostly about that.

1. **Upright or chest**: Uprights are easier to organize and access and lose more cold air when opened. Chest freezers hold temperature better during opening and door events and are harder to search. For frequently accessed inventories the upright usually wins on the time spent with the door open.
2. **Heat rejection and the room**: These units pump a great deal of heat into the room, and a row of them will defeat ordinary air conditioning. Establish the heat load and the room's capacity before ordering, because a warm room shortens compressor life and raises the failure rate of everything in the row.
3. **Power, circuits and backup**: Check the circuit rating and whether the unit sits on backup power. Decide deliberately which freezers are on generator or uninterruptible supply, because in a real outage the answer is whichever ones somebody thought about in advance.
4. **Alarms that reach a person**: An independent monitoring system that alerts a named person by a route they will actually receive, with a documented escalation, is the part that saves samples. A freezer's own beeper in an empty building at the weekend saves nothing.
5. **The written failure plan**: Where samples go, who moves them, who has keys, where dry ice comes from at two in the morning, and what spare capacity exists. Keep the plan where somebody who is not you can find it, and keep enough spare capacity in the building to take one freezer's contents.

**Upright or chest ultra-low freezer?**

Uprights are easier to organize and access but lose more cold air on opening; chest units hold temperature better and are harder to search. For frequently accessed inventories the upright usually wins overall.

**What should I check about the room?**

Heat rejection and the room's cooling capacity, the circuit rating, and whether the unit will sit on backup power. A row of freezers in an under-cooled room shortens the life of every unit in it.

**What kind of alarm do I need?**

Independent monitoring that alerts a named person by a route they will actually receive, with documented escalation. A built-in beeper in an empty building at the weekend protects nothing.

**What should the failure plan say?**

Where samples go, who moves them, who has access, where dry ice comes from out of hours, and what spare capacity exists. Keep it where someone other than you can find it.

**How should I compare ultra low freezers?**

On recovery after a door opening, performance in a warm room and behavior on a single compressor failure, not on the minimum temperature. Then on alarms, monitoring and whether backup injection is fitted.

**When do samples need cryogenic sample storage rather than a freezer?**

When viability has to survive years, as for cells intended to be revived. Mechanical freezers are too warm for that. Store in vapour phase rather than immersed, with an oxygen depletion monitor in the room.

## Equipment repair

Every instrument eventually fails, and the decision then is a three-way one: manufacturer service, an independent engineer, or replacement. The right answer depends on parts availability, whether software and diagnostics are locked to the manufacturer, and what requalification the instrument needs afterwards. This page sets out how to make that decision quickly rather than during a breakdown.

1. **Establish parts and software access first**: Some manufacturers restrict parts, diagnostic software or calibration routines to their own engineers, which effectively removes the third party option whatever their competence. Find this out for your major instruments before anything breaks.
2. **Manufacturer service, and what it buys**: Original parts, current firmware, the manufacturer's calibration procedure and a documented service record. It costs more and it is the straightforward choice for anything under warranty, anything qualified in a regulated environment, and anything where the service record matters.
3. **Independent engineers, and when they are right**: For mechanically simple, out-of-support or elderly instruments, a good independent engineer is frequently faster and much cheaper. Ask what parts they use, what calibration they perform, what documentation they issue and what happens to any remaining manufacturer support.
4. **Requalification after repair**: A repair that touches a measuring path means the instrument needs verifying, and in a regulated environment requalifying, before it is used again. Decide who does that and cost it into the repair rather than discovering it afterwards.
5. **The replace decision**: Repair stops being sensible when parts are unavailable, when software support has ended, when the repair approaches a meaningful fraction of replacement cost, or when downtime is costing more than the instrument. Write the threshold down in advance so the decision is not taken under pressure.

**Manufacturer or third party laboratory instrument repair?**

Manufacturer for anything under warranty, qualified in a regulated environment, or where parts and diagnostic software are restricted to them. Independent engineers are frequently faster and cheaper for mechanically simple or out-of-support instruments.

**Does an instrument need requalifying after repair?**

If the repair touched a measuring path, it needs verifying before use, and requalifying in a regulated environment. Decide who does that and include it in the repair cost.

**When should I replace rather than call a lab equipment repair service?**

When parts are unavailable, software support has ended, the repair approaches a meaningful fraction of replacement cost, or downtime is costing more than the instrument. Set the threshold before the failure, not during it.

**How do I reduce breakdowns?**

Preventive maintenance, keeping instruments inside their environmental limits, training users, and holding a small stock of the parts that stop work. Most failures are a consumable a scheduled visit would have caught.

## Lab automation

Automation projects overrun for a consistent reason: the laboratory automates the step that looked most tedious rather than the step that constrains throughput, and the constraint moves somewhere else. The cheapest automation project is the one that starts by measuring where work actually queues. This page covers choosing what to automate, in what order, and what the vendors in this market are actually selling.

1. **Measure where work queues**: Walk the process and find where samples wait, not where people complain. The constraint is frequently a manual data step, a plate sealing step or a review, rather than the pipetting everybody notices. Automating anything other than the constraint moves the queue rather than shortening it.
2. **Islands before integration**: Individual automated stations with a person moving plates between them are cheaper, more robust and easier to reconfigure than an integrated workcell. Integrate only where the manual transfer is genuinely the constraint, because every integration multiplies the failure modes.
3. **Sample identity must survive the whole path**: Barcodes, plate maps and their link to the system of record have to hold across every step. An automated process that cannot say which well came from which sample has automated the wrong thing, and this is the most common serious defect in these projects.
4. **Staff for it explicitly**: Automation replaces manual work with configuration, maintenance and troubleshooting work, and that work needs somebody whose job it is. Laboratories that automate without assigning the role end up with expensive instruments used manually.
5. **Plan the failure modes**: Decide what happens when a run fails halfway: whether samples are recoverable, whether the batch is lost, and how a partial run is recorded. Automation concentrates risk into fewer, larger events, and unplanned failure handling is where confidence is lost.

**What should a laboratory automate first?**

The step where work actually queues, which is frequently a data or review step rather than the pipetting. Automating anything else moves the constraint rather than removing it.

**Should I buy an integrated workcell?**

Usually not first. Individual automated stations with a person moving plates are cheaper, more robust and easier to reconfigure. Integrate only where the manual transfer is genuinely the constraint.

**Why do automation projects overrun?**

Because the wrong step was automated, because sample identity was not designed to survive the whole path, or because nobody was assigned to configure and maintain the system. All three are decided before purchase.

**Does automation reduce staffing?**

It changes it. Manual work becomes configuration, maintenance and troubleshooting work, and that needs someone whose job it is. Laboratories that skip this end up running automated instruments by hand.

**When is automation worth it?**

When a protocol is stable, repeated and worth fixing for a year. A protocol still being optimized will be reprogrammed constantly, and the automation then slows the work down.

**What is the hidden cost of liquid handling platforms?**

Integration: labware definitions matched to your consumables, scheduling, instrument integration, error handling and a path into your data system. A change of plate supplier can mean re-teaching every definition.

## CO2 incubators

A carbon dioxide incubator holds temperature, gas concentration and humidity around cells for weeks at a time, and the two things that decide whether it does that well are the sensor type and how quickly conditions recover after somebody opens the door. Contamination, when it happens, is usually traced to practice rather than to the machine, but the machine decides how painful recovery is. This page covers both.

1. **Sensor type decides behavior with humidity**: Infrared sensors measure carbon dioxide directly and are unaffected by humidity and temperature, so they recover quickly and do not drift after a door opening. Thermal conductivity sensors are cheaper and are affected by both, which means longer recovery and a longer stabilisation after decontamination.
2. **Recovery time after a door opening**: Cells experience the incubator as a series of disturbances, and a shared unit opened many times a day spends a meaningful part of its life recovering. Ask for recovery times for temperature and gas rather than only the stability specification, since the second is measured with the door shut.
3. **Jacket type and stability**: Water-jacketed units hold temperature longer through a power failure and are heavy and slow to heat. Air-jacketed units heat and recover faster and are lighter. Direct heat with a fan recovers fastest and moves air, which has consequences for evaporation and contamination spread.
4. **Decontamination cycle**: A built-in high temperature decontamination cycle removes the contamination problem thoroughly at the cost of a long out-of-service period. Cycles vary from several hours to most of a day, and how often you can afford that determines how much the feature is worth.
5. **Copper surfaces and filtration**: Copper interiors and high efficiency filtration both reduce contamination, and neither substitutes for aseptic technique. Treat them as useful margin rather than as a solution, since the usual source of contamination is what goes into the incubator.

**Infrared or thermal conductivity CO2 sensor?**

Infrared measures carbon dioxide directly and is unaffected by humidity and temperature, so it recovers faster and drifts less. Thermal conductivity is cheaper and is affected by both, giving longer recovery and stabilisation.

**Water jacketed CO2 incubator or air jacketed CO2 incubator?**

Water jackets hold temperature longer through a power failure and are heavy and slow. Air jacketed and direct heat units recover faster and weigh less. Recovery speed usually matters more day to day than power failure performance.

**How do I stop incubator contamination?**

Mostly through practice: aseptic technique, a weekly wipe-down, clean water with the recommended additive in the tray, and a well-maintained cabinet upstream. Copper surfaces and filtration add margin rather than a solution.

**What monitoring should I add?**

Independent temperature and gas monitoring with an alarm that reaches a person, plus a changeover regulator or cylinder alarm on the gas supply. Running out of carbon dioxide at a weekend is the commonest avoidable failure.

**Water jacket or air jacket?**

Air jacket for fast recovery with frequent door openings; water jacket for thermal mass that holds temperature through a power cut. They fail in opposite directions, so choose on which risk matters more to you.

## Flow cytometry antibodies

Building a flow panel is a matching problem: the brightest fluorochromes must go on the dimmest antigens, and every pairing has to be checked against what the instrument can detect and what will spill into neighbouring channels. Buy the antibodies before that matching is done and you will own a set of reagents that cannot be run together. This page covers the matching and the titration nobody has time for and everybody needs.

1. **Brightest fluorochrome on the dimmest antigen**: Resolution depends on the separation between positive and negative, so a dim antigen needs a bright fluorochrome to be seen at all, and a highly expressed antigen can afford a dim one. Getting this backwards is the single most common panel design error.
2. **Check the instrument configuration first**: A fluorochrome is only usable if the instrument has a laser to excite it and a detector to read it. Get the configuration and build the panel against it, because a beautiful panel on paper for an instrument that cannot read half of it is wasted money.
3. **Spillover and where it lands**: Emission spectra overlap, and the spillover from a bright fluorochrome degrades resolution in the channels it reaches. Avoid putting two bright dyes with heavy mutual spillover on two dim, co-expressed antigens, and expect compensation to cost resolution rather than restore it.
4. **Clone matters as much as the target**: Different clones against one antigen recognize different epitopes, and a fixation or permeabilisation step can destroy the epitope one clone needs. Choose the clone from published use in your application, not simply from the target name.
5. **Titrate every antibody**: The manufacturer's suggested volume is a starting point on their cells in their buffer. Titrating on your own cells finds the concentration giving the best separation, saves reagent, and reduces background. This is the step that most improves a panel and the one most often skipped.

**How do I match fluorochromes to antigens?**

Put the brightest fluorochromes on the dimmest antigens, because resolution depends on separating positive from negative. Check every choice against the instrument's lasers and detectors before ordering.

**Do I need to titrate every FACS antibody?**

Yes. The manufacturer's suggested volume was determined on their cells in their buffer. Titrating on yours improves separation, reduces background and saves reagent.

**Why does the clone of a flow cytometry antibody matter?**

Different clones recognize different epitopes on the same antigen, and fixation or permeabilisation can destroy the epitope a given clone needs. Choose the clone from published use in your application.

**Which controls are essential?**

Single stain controls for compensation, a viability dye to exclude dead cells that bind non-specifically, and fluorescence minus one controls wherever a gate boundary is not obvious.

## Library prep automation

Library preparation is repetitive, error-prone and long, which makes it an obvious automation candidate, and the obvious candidates are where automation projects most often disappoint. The case rests on throughput and on consistency between samples, not on the tedium, and the thing that must be designed first is how sample identity and index assignment survive every plate transfer. This page covers when the numbers work and what has to be right.

1. **Throughput is the threshold**: Below a steady flow of plates, a good technician is faster, cheaper and more adaptable than a platform that needs setup and maintenance for each run. Above it, automation wins on consistency as much as on hands-on time. Measure your actual sustained throughput rather than your peak.
2. **Consistency is the real gain**: Automated preparation reduces variation between samples and between operators, which shows up as more even coverage and fewer failed libraries. For anything quantitative or for long studies, this is worth more than the time saved.
3. **Index assignment must be traceable**: Every sample gets an index, and a mis-assigned index produces data attributed to the wrong sample, which is worse than a failed run because nothing looks wrong. The plate map, the index set and the system of record must be linked without anybody typing anything.
4. **Kit and deck compatibility**: Automated protocols are validated by kit suppliers for particular platforms, and running an unvalidated combination means you validate it. Confirm which kits have supported scripts on the platform you are considering, and whether the whole protocol fits on the deck.
5. **Quantification and normalisation steps**: Library quantification and pooling to equal molarity are the steps that decide whether samples are evenly represented. Ask whether the platform performs them, since a workflow automated up to that point and manual afterwards keeps the bottleneck it set out to remove.

**At what throughput does automated library prep pay?**

Above a sustained flow of plates rather than an occasional peak. Below that, a good technician is faster, cheaper and more adaptable than a platform needing setup and maintenance per run.

**What is the main benefit?**

Consistency between samples and operators, which shows as more even coverage and fewer failed libraries. For quantitative work and long studies that is worth more than the hands-on time saved.

**How do I avoid index misassignment?**

Link the plate map, the index set and the system of record without manual transcription, and use unique dual indexes. A mis-assigned index produces data attributed to the wrong sample with nothing looking wrong.

**What should NGS library quantification show in an evaluation?**

Your own kit on your own samples over a full plate, including a difficult sample type, compared against your manual process on library yield, size distribution and coverage evenness.

## LIMS for clinical laboratories

A laboratory system in a healthcare setting carries obligations a research system does not: it holds identifiable patient data, it exchanges orders and results with clinical systems, and it operates inside an accreditation regime that will inspect it. Those three things, rather than the feature list, are what separate the products in this market. This page covers what to establish before a selection.

1. **Patient identity and the master index**: The system must resolve a patient reliably across orders, episodes and identifiers, and merge duplicates safely. This is harder than it sounds and it is where clinical systems differ most from research ones, which key everything to a sample.
2. **Order and result interfaces**: Orders arrive from and results return to electronic health records, and those interfaces are the project. Establish which standards the system supports, which are in production at a named comparable customer, and who builds and maintains each interface.
3. **Accreditation and the inspection**: The laboratory operates under an accreditation regime that inspects records, competency, quality control and corrective actions. Ask specifically how the system supports the evidence an inspector asks for rather than whether it is compliant, which is a claim rather than a capability.
4. **Quality control and proficiency testing**: Control rules, control charts, out-of-control handling and proficiency testing records are daily operational work in a clinical laboratory. A system that treats them as reports rather than as workflow will be worked around.
5. **Turnaround and the critical result**: The system must support turnaround monitoring and the escalation of a critical result to a named person with acknowledgement recorded. Confirm how that path works and what happens when acknowledgement does not arrive.

**How does a healthcare LIMS differ from a research one?**

It manages patient identity across episodes, exchanges orders and results with clinical systems, and produces the evidence an accreditation inspection asks for. Research systems key everything to a sample and lack all three.

**Can a research LIMS be extended for clinical use?**

Rarely well. Patient identity management, clinical interfaces and the accreditation evidence trail are substantial gaps, and extending across them usually costs more than buying a product built for the regime.

**What decides the project timeline?**

The interfaces with clinical systems. Establish which standards are supported, which are live at a named comparable customer, and who builds and maintains each one.

**What should I ask about accreditation support?**

Not whether the system is compliant, which is a claim, but how it produces the specific evidence an inspector asks for: records, competency, quality control and corrective actions.

## GMP mRNA manufacture

Moving messenger RNA from research to compliant manufacture is not the same process run more carefully. The raw materials change grade, the plasmid becomes a controlled starting material, every analytical method needs qualifying, and the whole thing is documented under a quality system. Organizations consistently under-estimate the raw material lead times and the analytical work. This page covers what actually changes.

1. **Raw material grades and their lead times**: Nucleotides, enzymes, capping reagents and lipids all exist in research and compliant grades, and the compliant grades have long and sometimes uncertain lead times. This is the item most likely to set your timeline and the one least visible in a project plan, so order early.
2. **The plasmid becomes a starting material**: Template plasmid manufactured under a quality system, with its own cell bank, its own release testing and its own documentation, sits upstream of everything. A program that has been using research plasmid discovers this as a several-month step.
3. **Enzymes and their origin**: Polymerase, capping enzymes and the nucleases used in processing must be of appropriate grade with documented origin, and animal-derived components raise questions you would rather not answer later. Establish the position on each one.
4. **Analytics qualified, not just available**: Identity, integrity, capping efficiency, poly-A tail, residual double-stranded RNA, residual template DNA, residual protein and endotoxin all need methods that are qualified for their purpose. Analytical work routinely lags and routinely becomes the critical path.
5. **Documentation and the quality system**: Batch records, change control, deviation handling, personnel qualification and facility controls apply to everything. This is why compliant manufacture costs what it does and why it cannot be applied retrospectively to material already made.

**What is the main difference between research and GMP mRNA?**

Raw material grades, a controlled plasmid starting material, qualified analytical methods and a documented quality system covering everything. It is not the same process run more carefully, and it cannot be applied retrospectively.

**What usually sets the timeline?**

Raw material lead times for compliant grade nucleotides, enzymes, capping reagents and lipids, and the analytical qualification work. Both are routinely under-estimated in project plans.

**Does the plasmid need to be GMP?**

It becomes a controlled starting material with its own cell bank, release testing and documentation. Programs running on research plasmid find this is a step of several months.

**When should the transition happen?**

At a point where the comparability exercise between old and new process is cheapest and best supported by data you already hold. Keep retained samples of research material, because comparability is measured against them.

## Plasmid synthesis

A plasmid synthesis service builds a construct from a sequence rather than from a template, which frees you from what is available to clone but puts every design decision on you. What the quote depends on is the sequence's difficulty and the vector, and what the usefulness depends on is verification. This page covers specifying the order and the point at which producing a lot of a plasmid becomes an entirely different purchase.

1. **Sequence difficulty drives the price**: High GC content, long repeats, homopolymer runs and strong secondary structure all push a construct into a higher difficulty tier and a higher per-base rate. Providers will assess a sequence before quoting, and a construct flagged as difficult is a warning about turnaround as much as about cost.
2. **Vector choice is the other half**: Delivery in the provider's standard vector is cheapest; a specific expression, viral or shuttle vector is a separate line and sometimes requires you to supply it. Decide the vector from the downstream experiment before the order rather than re-cloning afterwards.
3. **Codon optimisation is a choice with consequences**: Optimizing for an expression host usually raises yield and changes the nucleotide sequence, which matters if the sequence itself is the subject or if regulatory elements sit inside the coding region. Say explicitly whether you want the native sequence or an optimized one.
4. **Verification and what you receive**: Expect full sequence verification of the delivered construct with the trace, a map, and the plasmid at a stated quantity and purity. A construct delivered without a verified sequence is a construct you will verify yourself before trusting.
5. **Quantity, and the format it arrives in**: Research orders arrive as a small quantity of dried or suspended DNA suitable for transformation. If you need working quantities immediately, specify the amount and the purity grade, since transfection-grade and endotoxin-limited material are separate specifications.

**What does a plasmid synthesis service deliver?**

A sequence-verified construct in a specified vector, with the verification trace, a map and a stated quantity and purity. It builds from a sequence rather than from a template, so no source material is needed.

**Should I codon optimise?**

Usually yes for expression, since it raises yield, and no if the nucleotide sequence itself is the subject or if regulatory elements sit inside the coding region. Say which you want explicitly.

**Is plasmid DNA production the same order?**

No. Producing working quantities is a fermentation and purification job with its own specifications for endotoxin, supercoiled fraction and residual host material, frequently from a different supplier.

**Will a provider refuse a sequence?**

Providers screen orders against sequences of concern and may query or decline one. Institutional approval for recombinant nucleic acid work remains the ordering laboratory's responsibility.

## Secondary antibodies

The secondary antibody is chosen last and causes a surprising share of western blot problems, because the rules governing it are simple and easy to get wrong. It must be raised against the host species of the primary, it must not react with anything else on the blot, and it must be titrated. Most high-background blots are a secondary antibody used at the concentration on the bottle. This page covers the three rules and the label decision.

1. **Match the host of the primary, not the sample**: A secondary is described as anti-something raised in something: an anti-rabbit raised in goat detects a rabbit primary. Read it as a sentence and the confusion disappears. The secondary must recognize the primary's host species, and it must not be raised in the same species as your sample.
2. **Cross-adsorption where species could collide**: Cross-adsorbed secondaries have been passed over immobilised serum from other species to remove antibodies that would bind them. Use them for multiplexing, for samples from a species close to the secondary's host, and any time an unexplained band appears with the secondary alone.
3. **Enzymatic or fluorescent label**: Horseradish peroxidase with chemiluminescence is sensitive and has a limited dynamic range that saturates easily, which makes quantitation awkward. Fluorescent secondaries give a wider linear range and straightforward two-color multiplexing, and need an imager. For quantitative work the fluorescent route is usually the better answer.
4. **Titrate it, like the primary**: The dilution printed on the bottle is a starting point. Running a small dilution series on your own blot finds the concentration giving the best signal to background, saves reagent and removes most background problems at a stroke.
5. **Run the secondary-only control**: Incubate a blot with the secondary and no primary. Any band that appears is the secondary binding something directly, and knowing that before you interpret a result saves a great deal of argument later.

**How do I choose a secondary antibody western blot detection needs?**

It must be raised against the host species of your primary and must not be raised in the same species as your sample. Read the name as a sentence: an anti-rabbit raised in goat detects a rabbit primary.

**What does cross-adsorbed mean?**

The secondary has been passed over immobilised serum from other species to remove antibodies that would bind them. Use cross-adsorbed reagents for multiplexing and whenever species could collide on the blot.

**HRP or fluorescent secondary antibodies?**

Chemiluminescence with peroxidase is very sensitive and saturates easily, which makes quantitation awkward. Fluorescent secondaries give a wider linear range and easy two-color multiplexing, and need an imager.

**Why is my background so high?**

Most often the secondary used at the dilution on the bottle. Titrate it on your own blot, and run a secondary-only control to see whether it binds anything directly.

**Why do I see bands at the immunoglobulin chain positions?**

The secondary is binding immunoglobulin already present in the sample. Use a secondary adsorbed against that species, or a conformation-specific reagent, and run a secondary-only lane to confirm the source.

## Reporter cell lines

A reporter assay turns a promoter, a pathway or a receptor into light, and almost every disappointing screen fails for one of three reasons: the reporter was mismatched to the timescale, the line drifted, or nothing was normalised. This page covers the reporter choice, the assay design, and the validation a reporter line owes before it is used to make decisions.

1. **Choose the reporter for the timescale you need**: Firefly luciferase has a short protein half-life, which is what makes it responsive: signal tracks transcription over hours rather than days. A stabilised fluorescent protein does the opposite and reports accumulation. If the biology you care about switches on and off within a working day, a short-lived enzymatic reporter is the honest choice.
2. **Decide whether the readout should be secreted**: A secreted reporter is sampled from the medium, so the same well is read repeatedly and the culture survives the experiment. A lysed reporter gives a larger dynamic range from a single endpoint. Time courses argue for secretion; single-point screens rarely do.
3. **Normalise against something that is not your pathway**: A second constitutive reporter in the same well corrects for cell number, transfection efficiency and pipetting. Choose a normaliser whose promoter is genuinely indifferent to your treatment, because a normaliser that responds turns a real effect into no effect.
4. **Make the line clonal, then prove it is stable**: A pool drifts. Isolate clones, measure the induction window on each, and keep the clone with the best window rather than the brightest baseline. Then measure that window again after ten and twenty passages before the line is trusted with a screen.
5. **Fix the controls before the first plate**: A vehicle control, a saturating positive control and an untransduced well belong on every plate. Without a saturating control there is no way to express a result as a fraction of maximum, and comparisons between plates stop being meaningful.

**What does fluc mean on a reporter construct?**

fluc is firefly luciferase, the most common enzymatic reporter. Its short half-life is the point: signal falls when transcription stops, so the assay reports a rate rather than a total.

**Do I need a dual luciferase reporter assay?**

You need a normaliser. A second, constitutive luminescent reporter read in the same well is the usual way to get one, and it corrects for cell number and handling in a way an adjacent well cannot.

**How many clones should a reporter cell line screen start from?**

Enough that you can afford to discard most of them. Induction window varies widely between clones from the same pool, and the first clone that grows well is rarely the one with the best window.

**Why did the assay window shrink after a few months?**

Silencing of the integrated construct, or outgrowth of a subclone with weaker expression. Both are why the induction window is re-measured at a fixed passage number and why frozen early-passage stock matters.

## Condenser and illumination

More resolution is lost to a badly set condenser than to any objective a laboratory is likely to buy. The condenser fixes the angle of light entering the specimen, and that angle, not the magnification printed on the turret, is what decides whether fine detail survives. This page covers how the illumination train is set, what to check on a used stand, and when brightfield stops being the technique for the sample.

1. **Match the condenser numerical aperture to the objective**: A condenser whose numerical aperture is below the objective's caps the resolution of the whole system at the lower of the two. An oil objective above a dry condenser is the most common way a laboratory pays for a lens it cannot use.
2. **Set Koehler illumination, every time the objective changes**: Focus the field diaphragm image on the specimen plane, center it, then open it just past the field of view. This gives even illumination and puts the lamp filament out of the image. It takes under a minute and is the single largest improvement available to most instruments.
3. **Set the aperture diaphragm by contrast, not by brightness**: Closing the aperture diaphragm raises contrast and destroys resolution; opening it does the reverse. Around two thirds to four fifths of the objective's aperture is the usual compromise. Never use it as a brightness control, which is what the lamp is for.
4. **Check the condenser height is actually adjustable**: On cheap stands the condenser is fixed or has no centring screws, which means Koehler illumination cannot be set at all. Confirm both the rack focus and the centring screws before buying, because no amount of objective quality compensates.
5. **Decide whether you need a swing-out top lens**: At low magnification a full condenser cannot illuminate the whole field. A swing-out top element fixes it. Without one, a 2x or 4x view will be vignetted and no adjustment will recover the corners.

**What does a condenser do on a microscope?**

It gathers light and delivers it to the specimen at a controlled range of angles. The range of angles, set by the aperture diaphragm and limited by the condenser numerical aperture, is what determines the resolution the objective can deliver.

**Why does my brightfield microscope image look flat?**

Usually the aperture diaphragm is wide open, or the field diaphragm was never focused on the specimen plane. Setting Koehler illumination and stopping the aperture back to roughly three quarters of the objective aperture fixes most of it.

**When is brightfield microscopy the wrong technique?**

When the specimen is unstained and transparent. Live cells in culture have almost no absorption contrast, which is what phase contrast and differential interference contrast exist to solve.

**Does immersion oil go under the condenser too?**

On a condenser rated above 1.0 numerical aperture, yes, and skipping it silently caps the system at about 0.95. Most laboratories never do it, which is why most high-aperture objectives are run below specification.

## Microscope slides

Slides look like the least consequential thing a laboratory buys, and then a batch arrives with a thickness tolerance that defeats an oil objective or a coating that lets every section float off in antigen retrieval. This page covers the three specifications that actually matter, when a charged or coated slide earns its price, and how slides and finished sections should be stored.

1. **Fix the substrate thickness and the flatness**: A standard slide is about one millimetre thick on a 25 by 75 millimetre footprint. Cheap slides vary within the box, and on an automated stage that variation shows up as a focus hunt on every field. For scanning work, specify the tolerance rather than assuming it.
2. **Match the coverslip to the objective, not to the budget**: Almost every dry objective above 20x is corrected for a 0.17 millimetre coverslip, which is what No. 1.5 means. No. 1 glass is thinner and introduces spherical aberration that no focus adjustment removes. For high numerical aperture work, specify 1.5H, which is the tightly toleranced grade.
3. **Choose the surface treatment by what will be done to the section**: Plain glass holds a smear. Charged slides hold sections through routine staining. Silane coated slides hold them through heat induced epitope retrieval, which is the step that lifts sections off everything else. Buy the treatment the protocol demands and no more.
4. **Decide on frosting and where the label goes**: A frosted end takes pencil and solvent resistant ink; a printable white end takes a laboratory printer. If sections will be tracked through a laboratory information system, a two dimensional barcode printed on the slide is worth the extra cost from the first day rather than the second year.
5. **Check cleanliness rather than assuming it**: Precleaned is a claim, not a specification. Residue shows up as uneven coating and as debris under fluorescence. A quick look at a blank slide under the instrument you will actually use is the only test that means anything.

**What thickness coverslip should I use?**

No. 1.5, which is nominally 0.17 millimetres, because that is what dry objectives above 20x are corrected for. No. 1 glass is thinner and costs you resolution that cannot be focused back.

**When do I need charged microscope slides?**

Whenever a section must survive aqueous steps. Charged glass holds tissue through routine staining, and a silane coating is what holds it through heat induced epitope retrieval.

**Does a microscope slide box need to be archival?**

If the sections are a record, yes. Cardboard folders absorb moisture and slides stick; a rigid microscope slide case with separated slots and a desiccant option is what keeps a stained section readable years later.

**Are all 25 by 75 millimetre slides interchangeable?**

On footprint yes, on thickness and flatness no. Automated stages and slide scanners are the first place a loose thickness tolerance shows up, usually as a focus problem blamed on the instrument.

## Top loading balances

A laboratory that buys only analytical balances ends up weighing kilogram quantities on an instrument designed for milligrams, and a laboratory that buys only top loaders eventually needs four decimal places it does not have. The two classes answer different questions. This page covers where each belongs, what the specification numbers mean in practice, and what a moisture balance is for.

1. **Start from the smallest quantity you will weigh, not the largest**: Readability has to be a small fraction of the smallest mass you care about. Weighing fifty milligrams on an instrument reading to ten milligrams gives a result with two significant figures, whatever the capacity above it.
2. **Then check the capacity you actually need**: Capacity and readability trade against each other on every instrument. A top loading balance covers roughly the range from a gram to several kilograms; below about a hundred milligrams of required precision the answer is an analytical instrument in a draught shield.
3. **Decide whether a draught shield is required**: Air movement dominates below about one milligram of readability, which is why analytical balances are enclosed. A top loader reading to ten or a hundred milligrams generally does not need one, and one sited beside a fume hood sash or an air conditioning outlet may still.
4. **Fix the calibration arrangement before purchase**: Internal calibration weights make a daily check trivial and are worth the premium on any instrument in regular use. External calibration needs a weight set of the right class, stored properly and itself recalibrated, which is a recurring cost people forget to quote.
5. **Site it before you buy it**: A solid bench away from doors, vibration and direct sun is worth more than a better specification on a bad bench. Balances returned as faulty are very often balances sited next to a centrifuge.

**What is the difference between a top loading balance and an analytical lab balance?**

Readability and enclosure. Analytical instruments read to 0.1 milligram or finer and are enclosed against draughts; a top loading balance reads coarser, carries more, and is open for easy access.

**What does a moisture balance measure?**

Mass loss on heating. It weighs a sample, heats it under a controlled program, and reports the percentage lost. It measures volatiles, not water specifically, which matters for anything that decomposes.

**How often should a balance be calibrated?**

Verified against a check weight at the frequency your quality system sets, usually each day of use, and calibrated by an accredited provider annually. The daily check is what catches a problem; the annual certificate is what documents it.

**Does balance lab equipment need to be on a dedicated bench?**

For analytical work, effectively yes: a stone or heavy bench away from traffic and airflow. For a top loader, a stable bench away from vibration sources is usually enough.

## Triple quadrupole MS

A triple quadrupole is bought for one job: quantifying known compounds at low concentration in a difficult matrix, reproducibly, for years. Everything in its specification serves that job, and comparing it against instruments built for discovery is how laboratories end up with the wrong tool. This page covers what the architecture buys, which numbers matter, and where the money goes.

1. **Confirm the job is targeted quantitation**: A triple quad selects a precursor, fragments it, and selects a product ion. That chain rejects matrix extremely well and is why detection limits are low. It cannot tell you what an unknown is, which is what a high resolution instrument is for.
2. **Count the transitions and the dwell time you need**: Each compound needs at least two transitions, one to quantify and one to confirm. Multiply by the number of analytes and divide the chromatographic peak width by the total, and you have the dwell time per transition. Too many analytes on a narrow peak is how sensitivity quietly disappears.
3. **Match the ion source to the chemistry**: Electrospray suits polar and ionisable compounds; atmospheric pressure chemical ionisation suits less polar ones. A source that does not suit the analyte cannot be compensated for downstream, and a second source is a significant later cost.
4. **Decide the polarity switching requirement early**: Methods covering acids and bases in one run need fast polarity switching, and switching time eats the duty cycle. If the method is already fixed, count the switches into the timing budget before believing a sensitivity figure.
5. **Price the whole system, not the box**: The chromatograph, the source, the gas supply, the vacuum pumps, the service contract and the software licenses are most of the five year cost. A quotation for the mass spectrometer alone is not a price.

**What does a triple quadrupole mass spectrometer do that a single quadrupole cannot?**

It adds a collision cell and a second mass filter, so it selects a precursor, fragments it and then selects a product. That double selection removes matrix interference and is why quantitation limits are orders of magnitude lower.

**When is a single quadrupole mass spectrometer enough?**

Clean samples, high concentrations, and confirmation rather than trace quantitation. For a synthesis laboratory checking that the right compound came out of a reaction, a single quadrupole is often the right and much cheaper answer.

**What drives mass spectrometer price most?**

Sensitivity class and the source and software around it. Between two instruments of the same architecture, the difference is usually ion transfer efficiency and duty cycle, and those are the specifications to make a vendor demonstrate on your own samples.

**Should I ask for a demonstration on my own matrix?**

Always, and with your own extraction. Sensitivity quoted on a clean standard tells you about the instrument in isolation; the number you will live with is the one measured through your sample preparation.

**Which analyser should I buy?**

A triple quadrupole for quantifying known compounds in a complex matrix; a high resolution instrument for identifying unknowns; a plasma source instrument for elements. The question decides, and the inlet matters as much as the analyser.

**Should I own one or send samples out?**

Own one for a steady stream of similar samples with a person whose job it is. Send out for occasional work, for accredited results and where the method does not exist in house; the service fee buys the method and the interpretation too.

## Calibration by instrument

Calibration is sold as a single service and is really a different measurement on every instrument. A certificate that says an instrument passed, without saying what was measured and against what, documents a visit rather than a result. This page sets out what calibration actually tests on the instrument classes a laboratory calibrates most, and what the certificate has to say for the record to be worth keeping.

1. **Spectrophotometer: wavelength accuracy, photometric accuracy and stray light**: Wavelength accuracy is checked against known absorption lines, photometric accuracy against certified neutral density filters, and stray light against a cut-off solution. All three matter, and a certificate quoting only wavelength has tested the least consequential of them for most assays.
2. **Analytical balance: repeatability, linearity, corner load and eccentricity**: A balance calibration is a set of measurements with traceable weights, not a single adjustment. Repeatability and corner load are what catch a failing instrument; a single mass placed centrally almost never does.
3. **Autoclave: thermocouple mapping under load, not a gauge reading**: The chamber gauge tells you about the chamber. What matters is temperature and time at the coldest point of an actual load, measured with distributed thermocouples, with a biological indicator to confirm lethality. Anything less is a pressure test.
4. **Fume hood: face velocity, containment and the alarm**: Face velocity across the sash opening is the routine measurement; a containment test with a tracer gas is what actually demonstrates the hood works. The sash alarm and the monitor are part of the test, because a hood whose alarm does not sound is a hood nobody knows has failed.
5. **Pipette: gravimetric verification at three volumes**: Volumes at the top, middle and bottom of the range, with enough replicates to get a standard deviation, corrected for temperature, pressure and humidity. A single volume check is a functional test and should not be recorded as a calibration.

**What should a calibration certificate state?**

The measurements made, the reference standards used with their own traceability, the environmental conditions, the measurement uncertainty and the acceptance criteria applied. A certificate without acceptance criteria records numbers nobody can act on.

**Do spectrophotometer calibration services need to be accredited?**

If the result supports regulated or accredited testing, yes: accreditation to ISO/IEC 17025 is what makes the traceability chain defensible. For internal method development an in-house check against certified filters is often proportionate.

**What lab calibration equipment is worth owning?**

A weight set for daily balance checks, certified absorbance filters for the spectrophotometer, and a calibrated thermometer. These turn an annual visit into a program, because they let you find a drift the week it starts rather than the year it is discovered.

**How often is calibration required?**

At the interval your quality system justifies, reviewed against the instrument's actual drift history. Annual is a convention, not a rule, and an instrument that has failed a check twice should be on a shorter interval regardless of convention.

**What does autoclave calibration measure?**

Temperature and time at the coldest point of an actual load, measured with distributed thermocouples, with a biological indicator to confirm lethality. The chamber gauge only describes the chamber, so a calibration that reads the gauge is a pressure test rather than proof that a load was sterilised.

**What does fume hood calibration check?**

Face velocity across the sash opening as the routine measurement, a tracer gas containment test as the evidence that the hood actually contains, and the sash alarm and monitor, because a hood whose alarm does not sound is a hood nobody knows has failed.

## Environmental LIMS

An environmental laboratory fails an audit for reasons a research laboratory never encounters: a holding time exceeded by an hour, a batch without its matrix spike, a deliverable in a format the receiving agency rejects. A general purpose laboratory system handles none of these natively. This page covers what an environmental lims has to do differently and how to test a candidate against it.

1. **Holding times computed from collection, and enforced**: The clock starts when the sample is taken, not when it is logged. The system must hold the collection timestamp, compute the expiry per analyte and method, and warn before it passes. A system that stores a date without acting on it has not solved the problem.
2. **Chain of custody as a record, not a printout**: Every transfer, with who released, who received and when, held as data and reproducible as a signed document. Custody that exists only as a scanned paper form attached to a record is custody the system cannot check.
3. **Quality control batching built into the workflow**: Method blanks, laboratory control samples, matrix spikes and duplicates belong to the preparation batch and their frequency is set by the method. The system should build the batch with its QC and refuse to report a batch whose QC has not been evaluated.
4. **The electronic data deliverable in the formats you owe**: Different agencies and clients demand different deliverable formats, and producing them by hand is where laboratories lose their margin. Ask which formats ship as standard, which are configurable, and what a new one costs.
5. **Accreditation evidence the system can produce**: Under a national accreditation program the laboratory must show method scope, analyst competence and traceability on demand. A system that holds these alongside the results turns an assessment into a query; one that does not turns it into a fortnight.

**What makes an environmental lims different from a general laboratory system?**

Holding times computed from collection, chain of custody as structured data, method-driven QC batching, and electronic deliverables in the formats regulators accept. A general system can be configured towards these; an environmental product starts there.

**Does the system need to know the method?**

Yes, because holding time, preservation, QC frequency and reporting limits all follow from the method. A system where the method is a free text field cannot enforce any of them.

**How important is the electronic deliverable?**

It is often the deciding factor. A rejected deliverable means rework on data that was correct, and the laboratories that lose least time are the ones whose system generates the format directly from the result record.

**Should subcontracted work live in the same system?**

It should. Subcontracted analyses still carry holding times and custody, and a report that combines in-house and subcontracted results has to show which is which without somebody reconciling two systems by hand.

## LIMS reporting

Reporting is the part of a laboratory system most often bought without a specification, and then rebuilt. The word covers three different jobs with different requirements: a controlled document sent to a customer, a dashboard the laboratory runs on, and an export analysts query. This page separates them and covers what each one needs.

1. **The certificate of analysis is a controlled document**: It is versioned, approved by a named person, reproducible years later exactly as issued, and traceable to the results behind it. That makes it a records problem rather than a formatting problem, and it belongs inside the system of record.
2. **The operational dashboard is about turnaround, not results**: Backlog by test, samples approaching their due date, instrument availability and rework rate are what a laboratory manager acts on. These are aggregates that change hourly and do not need the same controls as a certificate.
3. **The analytical export belongs to whoever asks the questions**: Trending across months, method comparison and customer analysis are best done in a tool built for it, fed by a stable read-only view of the result data. Building this into the laboratory system usually produces a slow report nobody can change.
4. **Decide where the template lives**: Customers ask for their own certificate layouts. A system where a new layout is configuration is very different from one where it is a development ticket, and the difference compounds with every customer you win.
5. **Pin down the electronic signature route early**: Where a certificate is signed electronically, the signature mechanism, the audit trail and the reissue path all have to satisfy the electronic records rule. Retrofitting that after go-live is expensive; specifying it before is not.

**Should reporting sit inside the LIMS or in a separate tool?**

Controlled certificates inside; ad hoc analysis outside, on a read-only view. Trying to do exploratory analysis inside the system of record gives you slow reports and a change control process on every question.

**What does a lims reporting system need for regulated work?**

A reproducible document tied to the approved results, an audit trail covering issue and reissue, and electronic signatures that meet 21 CFR Part 11 where that applies. Reproducibility is the hard part: the report has to regenerate identically years later.

**How do I handle customer-specific certificate layouts?**

Treat layout as configuration and data as fixed. One result set, many templates. Laboratories that branch the data to suit a layout end up unable to answer what a result actually was.

**Why are the dashboard numbers different from the report?**

Almost always a timing or a status difference: a dashboard counts everything in progress, a report counts what has been approved. Define the status each number is filtered on and print it on the dashboard.

## UHPLC systems

An ultra high performance system is not a faster chromatograph with the same plumbing. Small particles need high pressure and, more importantly, a flow path narrow enough that the sharp peaks they produce survive the journey to the detector. Laboratories that upgrade the pump and keep the tubing get the pressure and not the resolution. This page covers what actually has to change.

1. **Start with the pressure the column needs, then add margin**: Sub-two-micron particles at useful flow rates generate back pressures that older systems cannot reach. Specify the pressure at the flow rate and viscosity you will actually run, not the headline maximum, and leave room for a column ageing over its life.
2. **Count the extra-column volume, every connection**: The gain from a narrow, efficient column is lost in tubing, fittings, the injector and the detector cell. Short lengths of narrow bore tubing, zero dead volume fittings and a small detector cell are not refinements, they are the difference between a working system and an expensive one.
3. **Check the detector acquisition rate against the peak width**: A peak a second or two wide needs enough data points across it to be integrated reliably. A detector sampling too slowly reports a smaller, broader peak and nobody can tell from the chromatogram that the instrument caused it.
4. **Plan the injector and the carryover specification**: Small volumes and high sensitivity make carryover visible. Ask for a measured carryover figure on a compound like yours, with the wash configuration that produced it, rather than a specification measured on something convenient.
5. **Decide whether micro hplc suits the sample better**: Where sample is scarce or sensitivity per mass injected matters more than speed, narrow columns at low flow give more signal from less material. It is a different trade from raw speed, and for precious samples usually the better one.

**What is the difference between HPLC chromatography and an ultra high performance liquid chromatography system?**

Particle size and everything that follows from it. Sub-two-micron packing gives sharper peaks, needs much higher pressure, and needs a flow path with very little dispersion or the sharpness never reaches the detector.

**Can I transfer an existing HPLC method to a UHPLC system?**

Usually yes, by scaling the column dimensions, flow rate, injection volume and gradient together. Where the method is a registered one, the scaling has to stay within the allowed adjustment ranges or it is a change requiring revalidation.

**Why did my peaks get worse after the upgrade?**

Extra-column dispersion, almost always. Long or wide tubing between injector, column and detector, or a large detector flow cell, will broaden narrow peaks regardless of how good the column is.

**Is micro hplc the same thing?**

No. Micro scale reduces column diameter and flow rate to get more sensitivity from limited sample. Ultra high performance reduces particle size to get resolution and speed. They can be combined, and they answer different problems.

**Do I need a high pressure chromatography pump?**

Only if throughput or resolution is the constraint and the whole system can be low dispersion. A high pressure pump in a conventional stack delivers a fraction of the benefit, because tubing, cell volume and data rate then limit the result.

## HPLC pumps and hardware

The pump decides retention time reproducibility, and retention time reproducibility decides whether a method survives contact with a second laboratory. Most pump problems announce themselves in the chromatogram long before an error appears on the screen. This page covers choosing a pump, recognising the wear patterns, and the hardware between the pump and the column that quietly undoes its work.

1. **Choose binary or quaternary by what the method needs**: A binary system mixes at high pressure and generally has a smaller dwell volume and better gradient fidelity. A quaternary mixes at low pressure from four lines and buys convenience in method development. If gradient methods will be transferred between instruments, the dwell volume difference matters more than the convenience.
2. **Know your dwell volume and write it in the method**: Dwell volume shifts every gradient method's retention times, and it differs between instruments. A method transferred without stating the dwell volume it was developed on will not reproduce, and the laboratory receiving it will blame the column.
3. **Read seal wear in the baseline**: Rising pressure ripple, a drifting baseline and slow leaks at the pump head are seal wear. Buffers crystallise and accelerate it, which is why a wash line behind the piston is worth having on any instrument running salts.
4. **Read check valve failure in the pressure trace**: Erratic pressure with a regular period, and retention times that wander, usually mean a check valve is not seating. Cleaning sometimes works; replacing is the reliable answer and check valves are consumables rather than parts.
5. **Match hplc column hardware to the pressure class**: Fittings are not universal. Ferrule geometry and stud length differ between manufacturers, and a fitting seated at the wrong depth leaves a void at the column inlet that shows as a split or tailing peak on every injection.

**Binary pump HPLC system or quaternary?**

Binary for gradient fidelity, low dwell volume and method transfer. Quaternary for flexibility in development and for blending buffers on the fly. Many laboratories run both for exactly those reasons.

**How often should pump seals be replaced?**

On a schedule derived from your own history, not from the manual alone. A system running aqueous buffers all day wears seals far faster than one running organic mobile phases, and the pressure trace tells you when the interval is wrong.

**Why do retention times drift between two identical systems?**

Dwell volume and mixing behavior. Two instruments of the same model with different mixer configurations produce different gradients at the column head, which is why the mixer volume belongs in the method record.

**Does hplc column hardware really differ between brands?**

Yes, in ferrule geometry and the depth the tubing seats. Using one brand's fitting in another's column is the usual cause of a dead volume at the inlet, and the peak shape it produces is often blamed on the column packing.

## Loading buffer

Sample buffer is four reagents and a dye, and almost every component is there for a reason that becomes obvious the first time it is left out. It is also where several common blot failures begin: a membrane protein aggregated by boiling, a reducing agent that oxidised in the freezer, a sample too dilute to load. This page covers what each component does and how to prepare and store the buffer.

1. **SDS: the uniform negative charge**: Sodium dodecyl sulphate coats the unfolded polypeptide with negative charge roughly in proportion to length, which is what makes migration a function of size. Too little and the separation is not by size; far too much and the front distorts.
2. **Tris at pH 6.8: the stacking buffer**: The sample buffer matches the stacking gel so the sample concentrates into a sharp band before it enters the resolving gel. A sample buffer at the wrong pH gives broad, smeared bands that look like a transfer problem.
3. **Glycerol: density so the sample stays in the well**: Glycerol makes the sample dense enough to layer under the running buffer. Concentrated stock is viscous, which is the practical reason a 5x buffer is harder to pipette accurately than a 2x one.
4. **Bromophenol blue: a visible front**: The dye tracks the ion front so you can see the run and stop it before the smallest proteins leave the gel. It is also the fastest way to spot a well that did not load.
5. **A reducing agent, added fresh**: A thiol reagent breaks disulphide bonds so the protein runs as a single chain. It oxidises in storage, which is why it is added to the aliquot on the day rather than kept in the stock. A stored complete buffer is the commonest reason a sample stops reducing properly.

**How do I make a 5x western blot loading buffer?**

Tris-HCl at pH 6.8, SDS, glycerol and bromophenol blue as a 5x stock, stored in aliquots at minus twenty. The reducing agent goes in on the day of use, because it oxidises in the freezer.

**Should I always boil samples?**

No. Ninety-five degrees for five minutes suits most soluble proteins. Multi-pass membrane proteins aggregate when boiled and often run better at thirty-seven or seventy degrees, which is a change worth testing before you conclude a protein is not expressed.

**Reducing or non-reducing?**

Reducing for most work. Non-reducing where the antibody recognizes a conformational epitope or where you want to see the assembled complex, in which case the sample must not be heated hard either.

**Why are my bands smiling or smeared?**

Usually too much protein in the well, or a sample buffer whose pH or salt content has drifted. High salt samples distort the field locally, and diluting into fresh buffer rather than adding concentrated stock to a salty sample fixes most of it.

**Is a Laemmli buffer western blot stock the same as this sample buffer?**

Yes. A Laemmli buffer is the Tris, SDS, glycerol and bromophenol blue stock described above, named after the paper the discontinuous system comes from. The reducing agent is the only part that is not in the stored stock, and that is a storage decision rather than a difference of recipe.

## Protein ladders

A ladder answers two questions and people usually only ask one of them. It tells you roughly how big a band is, and it tells you whether the gel ran and the transfer worked. Those two jobs want different products, and the apparent weights printed on the vial are calibrated to a particular gel system rather than being absolute. This page covers choosing one and reading it honestly.

1. **Prestained, for watching the run and the transfer**: Dye-coupled proteins are visible in the gel and on the membrane, so you can see the separation develop and confirm the transfer before any antibody is used. The dye coupling shifts mobility, which is why the stated weights are approximate.
2. **Unstained, where the size estimate has to be as good as it can be**: Without dye the protein runs closer to its true mobility, so sizing is better. It is invisible until stained, which makes it useless for monitoring and means the lane has to be stained or the ladder detected separately.
3. **Dual color, to anchor the eye**: Reference bands in a contrasting color at known positions let you read the ladder at a glance without counting from the top. On a blot photographed in a hurry that is worth more than it sounds.
4. **Match the range to the target, not to the catalog**: A ladder spanning ten to two hundred and fifty kilodaltons is fine for a forty kilodalton protein and useless for resolving two bands ten kilodaltons apart at the bottom of the gel. Pick a ladder whose bands bracket the target closely.
5. **Handle it like a reagent that degrades**: Ladders tolerate a limited number of freeze-thaw cycles before bands fade and smear. Aliquot on arrival, and do not heat a prestained ladder harder than the instructions say: overheating is the usual cause of a ladder that suddenly runs as a smear.

**Prestained protein ladder or unstained protein ladder?**

Prestained to monitor the run and the transfer, unstained when the molecular weight estimate matters. Many laboratories run both, on opposite sides of the gel.

**Why do the ladder weights not match the literature?**

Because apparent molecular weight depends on the gel system, and prestained ladders are shifted further by the dye. The values printed on the vial are calibrated for a specific gel chemistry, and they change if you change it.

**Can I use a western blot protein ladder for quantitation?**

Only if it is sold as a quantified ladder with a stated mass per band. Ordinary ladders are for position, not for amount, and treating band intensity as a standard curve will mislead you.

**How much ladder should I load?**

The volume on the datasheet, and no more. Overloading produces a bright smear that bleeds into the neighbouring lane, which is a recurring cause of a phantom band at the edge of the first sample.

## Polyclonal antibody production

A polyclonal antiserum is a population of antibodies against many epitopes, which makes it sensitive, tolerant of a denatured target, and finite. That finiteness is the decision people underweight: when the animal is gone, the reagent is gone. This page covers how a project is specified, what each stage delivers, and what has to be true before the serum is used for anything that matters.

1. **Choose the host by phylogenetic distance and volume needed**: The host must see the antigen as foreign, so a target conserved between your species and the host gives a poor response. Rabbits give a good response and a useful serum volume; larger hosts give far more serum for a reagent that will be used for years; mice give little and are usually a route to a monoclonal instead.
2. **Design the immunogen, and conjugate it if it is small**: A peptide under roughly ten kilodaltons will not raise a response alone and must be coupled to a carrier protein. Choose a peptide from an exposed, non-conserved region, and check it against the proteome so the serum does not recognize half the lysate.
3. **Fix the schedule and the bleeds in the contract**: A standard program is a primary immunisation with boosts every two to three weeks and test bleeds through a seventy to ninety day cycle. Agree what happens if titre is low at the first test bleed, because that decision point is where projects either recover or drift.
4. **Decide the purification before the serum arrives**: Whole serum is cheapest and has the most background. Protein A or G gives total immunoglobulin. Antigen affinity purification gives the specific fraction and is what most applications actually need. The right answer depends on the application and it is much cheaper to specify than to add later.
5. **Validate against a negative that should be negative**: Preimmune serum from the same animal, and a sample lacking the target, are the two controls that separate a specific reagent from a hopeful one. Both belong in the deliverable rather than in a later experiment.

**How long does custom polyclonal antibody production take?**

Typically seventy to ninety days from immunisation to a final bleed, plus purification. Peptide synthesis and carrier conjugation add a few weeks at the front, and a poor first response can add a whole boost cycle.

**Why choose rabbit polyclonal antibody production?**

Rabbits raise a strong response to most mammalian targets, give a usable serum volume, and there is a very wide range of anti-rabbit secondary reagents available. For most laboratory applications it is the default for good reasons.

**Polyclonal or monoclonal?**

Polyclonal for sensitivity, for denatured targets and for immunoprecipitation, and where a finite supply is acceptable. Monoclonal where the reagent must be identical for years, where specificity has to be exact, or where the assay will be validated once and run indefinitely.

**Do I own the antigen and the serum?**

Only if the contract says so. Ownership of the immunogen, the serum, any residual and the right to resell are separate points, and providers differ. Settle them before work starts.

## Chromatin and histone antibodies

Chromatin immunoprecipitation is the application where a poor antibody does the most damage, because the output is a genome-wide map that looks plausible whatever was pulled down. A histone modification antibody has to distinguish one methyl group from two on the same residue, and many sold for the purpose do not. This page covers the evidence that separates them.

1. **Demand modification specificity data, not just a blot**: A western blot on bulk histones shows the antibody binds something the right size. What matters is whether it discriminates between neighbouring modification states, and the evidence for that is a peptide array or a dot blot against a panel of modified peptides.
2. **Run a peptide competition**: Pre-incubating the antibody with the free modified peptide should abolish the signal, and the unmodified peptide should not. This is a cheap experiment in your own hands and it is the fastest way to expose a reagent that binds the backbone rather than the mark.
3. **Use a genetic negative where one exists**: A knockout or a point mutant of the modified residue is the strongest control available. Where the biology allows it, an enzyme inhibitor that removes the mark gives much of the same information.
4. **Test the lot you will use, not the lot on the datasheet**: Polyclonal chromatin antibodies vary substantially between lots, and a publication validating lot A says little about lot B. Buy enough of one lot to finish the project, and record the lot number with every experiment.
5. **Include the input and a non-specific immunoglobulin**: Input normalises for chromatin preparation and sonication efficiency; an isotype-matched immunoglobulin control shows the background of the procedure itself. Neither is optional and both are routinely missing from published methods.

**What makes a chromatin antibody suitable for ChIP?**

It has to bind the epitope in fixed, fragmented chromatin rather than only on a denatured blot, and it has to discriminate the modification state. Ask for ChIP data with a genomic control region, plus peptide array specificity data.

**Why does lot variation matter so much for histone antibodies?**

Because most are polyclonal and the mark is a very small epitope. A new bleed is a new mixture of clones, and its discrimination between mono, di and tri methylation can differ substantially from the previous one.

**Are recombinant monoclonal chromatin antibodies better?**

For reproducibility, clearly yes: the sequence is fixed so the reagent is the same indefinitely. They still have to be validated for specificity, because a defined clone can be defined and wrong.

**What does a peptide competition control actually prove?**

That the signal depends on the epitope you think it does. It does not prove the antibody discriminates between related modifications, which is why the competition is run with both the modified and the unmodified peptide.

## IHC marker antibodies

Marker antibodies fail in tissue for reasons that have nothing to do with the antibody: the wrong retrieval, a fixation time nobody recorded, a blocking step that leaves endogenous peroxidase active. This page works through three widely used markers, what each one is actually reporting, and the method variables that decide whether a section is interpretable.

1. **Know what the marker labels, and what it does not**: CD31 labels endothelium and is used to read vessel density, and it is also present on platelets and some leukocytes. GFAP labels astrocytes and rises with reactive change rather than with cell number. Iba1 labels microglia and macrophages and does not distinguish resident from infiltrating cells. Each of these caveats has been over-read in print.
2. **Fix the fixation before blaming the antibody**: Formalin fixation time changes epitope availability substantially, and over-fixed archival material needs harder retrieval. Record fixation time with the block; a cohort fixed inconsistently cannot be compared however good the staining looks.
3. **Choose the retrieval condition empirically**: Heat induced retrieval in citrate near pH six and in a Tris or EDTA buffer near pH nine give different results for the same antibody. Test both on a known positive section rather than following the first protocol you find.
4. **Block what actually causes the background**: Endogenous peroxidase in blood-rich tissue, endogenous biotin in liver and kidney, and non-specific protein binding each need a different block. Adding more serum does not fix a peroxidase problem.
5. **Run the controls on the same slide run, every time**: A known positive tissue, a negative tissue and a no-primary control. Where a section is scored quantitatively, the positive control also fixes the exposure and threshold, and without it two runs are not comparable.

**What does a cd31 antibody stain?**

Endothelial cells, which is why it is the usual marker for vessel density. It also marks platelets and subsets of leukocytes, so a count based on it should be restricted to structures that look like vessels.

**Does an anti GFAP antibody count astrocytes?**

Not reliably. It labels the intermediate filament, which is strongly upregulated in reactive astrocytes, so a stronger signal usually means more reactive cells rather than more cells. A nuclear marker is needed alongside it for counting.

**What is an iba1 antibody used for?**

Labeling microglia and macrophages, including their fine processes, which makes it the standard marker for microglial morphology. It does not separate resident microglia from infiltrating monocytes on its own.

**Which retrieval buffer should I start with?**

Try citrate near pH six and a Tris-EDTA buffer near pH nine on the same known positive tissue. Datasheet recommendations are a starting point and often reflect one laboratory's fixation rather than yours.

**Does a fibroblast activation protein antibody identify tumour fibroblasts?**

It identifies activated fibroblasts, which appear in wound healing and fibrosis as well as in tumour stroma. Pair it with a lineage marker on the same section and a control tissue in which the state is absent.

## Antibody engineering

Engineering an antibody is a series of trade-offs made in a fixed order, and the expensive mistakes are the ones made late. Format decides pharmacokinetics and manufacturability; sequence decides specificity and immunogenicity; developability decides whether any of it survives scale-up. This page sets out the order and what each stage should produce.

1. **Choose the format from the job, not from the platform**: A full immunoglobulin gives long circulating half-life and effector function. A Fab removes effector function and keeps bivalency out of the picture. An scfv antibody is small, penetrates tissue well, is easy to display and express, and is cleared quickly and prone to aggregation. For a research reagent, an scFv is often ideal; for a therapeutic, rarely as the final format.
2. **Fix the sequence and make it recombinant early**: Recombinant monoclonal antibodies are expressed from a known sequence, so the reagent is identical indefinitely and can be reformatted at will. The hybridoma that produced the original clone will eventually drift or be lost, which is why sequencing it is worth doing while it is healthy.
3. **Humanise only as far as the application requires**: A research reagent does not need humanising. A therapeutic candidate does, and grafting the complementarity determining regions onto a human framework usually costs affinity that then has to be recovered by back-mutating framework residues one at a time.
4. **Affinity mature against the real requirement**: Higher affinity is not always better: for a tissue-penetrating agent very high affinity can trap the molecule at the periphery of a target. Define the affinity you need from the application before running the campaign.
5. **Screen developability before committing**: Thermal stability, aggregation propensity, expression titre, polyreactivity and the presence of chemical liabilities in the sequence can all be assessed on small quantities. A candidate that fails these at milligram scale will fail them at kilogram scale for much more money.

**What is an scfv antibody?**

A single chain variable fragment: the heavy and light variable domains joined by a peptide linker into one polypeptide. It keeps the binding site, loses the constant regions, and is small, easy to express and easy to display.

**Why convert a hybridoma monoclonal antibody to a recombinant monoclonal antibody?**

Permanence and control. Once the variable region sequence is known the reagent can be produced indefinitely, reformatted into any isotype or fragment, and produced without lot-to-lot variation.

**Do recombinant monoclonal antibodies still need validation?**

Yes. A defined sequence guarantees consistency, not correctness. A recombinant reagent that cross-reacts will cross-react identically forever, which is a reproducible error rather than an absent one.

**When should developability screening happen?**

As soon as there is a panel worth ranking, which is usually well before lead selection. Screens done after a lead is chosen tend to be used to justify it rather than to choose it.

## Conjugated antibodies

A directly conjugated antibody removes a secondary reagent, a species conflict and an incubation step, and introduces two new variables: how many dye molecules are attached, and whether the attachment has damaged the binding site. Neither is on the tube. This page covers how conjugates are specified, how to test one, and when to make your own.

1. **Read the degree of labeling, and ask for it if it is absent**: The number of dye molecules per antibody is what decides brightness, and past a point it decides quenching too. Roughly three to six is the usual working range for small organic dyes. Too few is dim; too many is dim as well, and also more prone to non-specific binding.
2. **Titrate on your own sample**: The dilution on the datasheet came from somebody else's cells, staining buffer and instrument. Run a dilution series and choose the concentration at the best separation between positive and negative, which is very often several times more dilute than recommended.
3. **Choose brightness to match antigen density**: Put the brightest fluorophores on the dimmest targets. A very bright dye on an abundant antigen spreads into neighbouring detectors and costs you resolution elsewhere in the panel for no gain.
4. **Check the conjugate has not damaged the antibody**: Compare the conjugate against the same clone detected indirectly on the same sample. A conjugate that stains noticeably worse has been over-labeled or labeled at a residue that matters, and no titration will recover it.
5. **Store it as a light-sensitive protein**: Conjugates degrade with light and with freeze-thaw far faster than unlabelled antibodies. Aliquot on arrival, keep them dark at four degrees unless the supplier says otherwise, and never freeze a tandem dye.

**What is degree of labelling?**

The average number of dye molecules attached per antibody. Around three to six is typical for small organic dyes; below that the conjugate is dim, above it self-quenching and non-specific binding both rise.

**Should I buy antibodies conjugated or conjugate my own?**

Buy for common clone and fluorophore combinations, where the supplier's quality control is worth more than the saving. Conjugate in house when the combination is not sold, when you need a specific ratio, or when a custom hapten is involved.

**Why is my directly conjugated antibody dimmer than the indirect stain?**

Indirect detection amplifies, because several secondaries bind each primary. A direct conjugate gives one antibody's worth of signal, which is cleaner and quantitatively better behaved, and genuinely dimmer.

**Can I freeze fluorophore conjugated antibodies?**

Check the supplier's instruction rather than the freezer convention. Many conjugates tolerate a single freeze in aliquots; tandem dyes generally do not and degrade into a signal that appears in the wrong detector.

## Antibody purification

Purification is where a good antibody is commonly damaged. Acid elution, aggregation on concentration and a final buffer chosen by habit account for most of the activity lost between a successful expression and a reagent that works. This page covers the routes, what each one actually separates, and the handling that preserves what you purified.

1. **Capture on protein A or protein G, chosen by species and subclass**: Protein A binds many immunoglobulin G subclasses well and some poorly; protein G has a broader subclass range and binds some species where protein A does not. Check both against the species and subclass you have rather than defaulting, because a poor match halves the yield before anything else happens.
2. **Use antigen affinity when specificity is the product**: Protein A or G gives total immunoglobulin, including everything the animal made against everything else. Passing it over immobilised antigen gives the specific fraction, which is what a polyclonal reagent usually needs and what a monoclonal does not.
3. **Neutralise the eluate immediately**: Affinity elution is at low pH, and time at low pH is what denatures and aggregates antibodies. Collect into a neutralising buffer already in the tube rather than neutralising afterwards, and keep the fractions small.
4. **Polish on size exclusion to remove aggregate**: Aggregate causes high background in every application and is invisible on a reducing gel. A size exclusion step both removes it and tells you how much there was, which is a quality measurement worth keeping for each preparation.
5. **Formulate for storage, not for the column**: The final buffer decides shelf life. A neutral buffered saline with a carrier protein and a cryoprotectant suits most reagents; sodium azide preserves but interferes with peroxidase detection, which is a conflict worth resolving before the aliquots are made.

**Protein A chromatography for antibody purification, or protein G?**

Decide from the species and subclass. Protein G has broader subclass coverage for several species; protein A gives cleaner elution for others. For mixed subclasses, a protein A/G resin avoids losing part of the preparation.

**How do I avoid aggregation during purification?**

Minimize time at low pH by neutralising into the collection tube, avoid concentrating beyond what the application needs, and polish on size exclusion. Concentration steps cause more aggregate than elution does in most preparations.

**Is antigen affinity purification always better?**

For a polyclonal serum, usually, because it removes the antibodies raised against everything else. It also loses low-affinity specific antibodies, which occasionally matters, and it needs enough immobilised antigen to be worth doing.

**What should the final protein purification buffer be?**

Whatever is compatible with the application and the storage plan. Phosphate buffered saline with a stabiliser is a common default; check azide compatibility with your detection chemistry and whether the reagent will be conjugated later, because carrier protein blocks amine coupling.

**Is general immunoglobulin capture good enough?**

For many routine uses, yes. It recovers all the animal's immunoglobulin, not only the antibodies against your target, so for blocking, immunoprecipitation or any conjugation, affinity purification against the immunogen is the cleaner and usually the better buy.

## Aseptic fill and finish

Fill and finish is the last operation and the one that most often delays a program, because contamination control is a facility property rather than a process parameter. The decisions that matter are made long before the first batch: the container, the closure, the line, and the evidence the site can already produce. This page covers how to scope it and what to ask a provider.

1. **Choose the container closure system first**: Vial, prefilled syringe or cartridge decides the line, the stoppering, the inspection and much of the stability work. Changing it later is effectively restarting. Decide it from the route of administration and the intended user, not from what the site already runs.
2. **Establish the product's tolerances before the line is selected**: Shear sensitivity, silicone oil interaction, tungsten residue from syringe manufacture and surface adsorption all attack biologics in ways that only appear in the final container. Small scale studies on these cost weeks; discovering them at engineering batch costs months.
3. **Ask for the site's media fill history, not its capability statement**: Three consecutive successful media fills at the scale and line you would use, with the interventions represented, and an acceptance criterion of no contaminated units. Ask what interventions were simulated, because a media fill without the hard interventions proves little.
4. **Read the environmental monitoring data**: Grade A monitoring results over a period, with excursion investigations and how they were closed, tell you more about a facility than any tour. A site unwilling to share trended data under confidentiality is telling you something.
5. **Fix container closure integrity testing early**: Deterministic methods such as helium leak or high voltage leak detection have largely replaced the dye ingress test, and the method has to be developed for your specific container. This is routinely left until stability is already running.

**What does aseptic fill and finish cover?**

Formulation of the final drug product, sterile filtration, filling into the primary container, stoppering and sealing, inspection, and labeling and packaging. It is the step where the product becomes the thing a patient receives.

**How do I choose a fill finish cdmo?**

On the container formats it runs routinely, its media fill and environmental monitoring record, its inspection capability, and slot availability. Capacity in eighteen months is not capacity, and the format it runs occasionally is the one that will cause problems.

**What changes with syringe filling machines?**

Prefilled syringes bring silicone lubrication, the plunger and the needle shield into the product contact picture, plus functional testing of break-loose and glide force. It is a materially larger validation exercise than vials.

**What is the acceptance criterion for a media fill?**

No contaminated units, with any positive triggering a full investigation. Three consecutive successful runs are the conventional qualification, with periodic requalification and any significant change requiring repetition.

## Bioanalytical services

Bioanalysis is bought as a per-sample price and delivered as a defensible dataset, and the gap between those two is where programs lose time. Validation status, matrix, species and the regulatory context all change what the same assay costs and what it is worth. This page covers how to specify the work and what to ask before placing it.

1. **Decide the regulatory context before the assay**: A discovery screen, a study supporting a regulatory submission and a clinical trial sample analysis are three different exercises with the same chromatogram. The context decides whether a fit for purpose qualification suffices or a full validation is required, and it decides the price.
2. **Fix matrix and species at the outset**: An assay validated in one species' plasma is not validated in another, and anticoagulant, haemolysis and lipaemia all change recovery. Supply real matrix from the intended species early so selectivity is established on it rather than on a surrogate.
3. **Agree the calibration range against the expected exposure**: A range that does not bracket the concentrations that occur means dilutions, repeats and delay. Model the expected profile before the range is set, and agree how out-of-range samples will be handled.
4. **Require stability data that matches the real sample journey**: Bench top, freeze-thaw, long term storage and whole blood stability have to cover the handling the samples will actually see, including the courier leg. Stability gaps found after collection cannot be fixed retrospectively.
5. **Confirm incurred sample reanalysis is planned**: Reanalysing a subset of real study samples is what demonstrates the method works on the samples rather than on spiked standards. It is an expectation in regulated work and a good idea in every other kind.

**What is included in bioanalytical services?**

Method development, validation or qualification, sample receipt and storage, analysis, data review and reporting. The price per sample generally excludes the validation, which is a separate and larger item.

**When do I need full validation rather than a qualification?**

When the data supports a regulatory submission or a clinical decision. Discovery work is usually served by a fit for purpose qualification, and upgrading later means revalidating rather than extending.

**What is a pharmacokinetic assay expected to deliver?**

Concentrations with defined accuracy and precision across a range that brackets the observed exposure, in the study matrix, with documented stability covering how the samples were handled from collection to analysis.

**How do I compare quotes from two bioanalytics services providers?**

Normalise the scope: validation extent, number of analytes, sample volume, turnaround, repeat policy, incurred sample reanalysis and reporting format. Per-sample prices for different scopes are not comparable and almost never are.

## Spectral flow cytometry

A spectral cytometer measures the whole emission signature of every event across many detectors and then solves for which fluorophores produced it. That changes what panels are possible and changes how they fail. The failures are quieter than compensation errors and easier to publish by accident. This page covers what the approach does and how to run it honestly.

1. **Understand what unmixing is solving**: Conventional cytometry assigns one detector to each fluorophore and corrects for spillover. A spectral instrument records the full emission signature and solves a system of equations for the contribution of each fluorophore. It can separate dyes with similar peaks but different shapes, which filter-based detection cannot.
2. **Record a reference signature for every fluorophore, on the right carrier**: Unmixing needs a pure signature per fluorophore, from a single stained control. The carrier matters: beads and cells give different signatures for the same dye, and mixing carriers between controls is the commonest cause of unmixing artefacts.
3. **Take autofluorescence seriously, because you now can**: An unstained control from the same sample type lets autofluorescence be treated as an additional signature and extracted. On tissue, macrophages and anything fixed, that is a substantial gain, and it requires an unstained control per sample type rather than one for the run.
4. **Check the unmixing before gating anything**: Look at the unmixed data for populations that have moved off-axis or acquired an impossible shape. Spectral unmixing errors look like biology, which is why the check is visual and deliberate rather than automatic.
5. **Design the panel by similarity, not by peak**: Two dyes with the same peak and different shapes can be resolved; two with very similar full spectra cannot, whatever their peaks. Use a similarity measure rather than a color chart, and put the most similar pair on the most distinct antigens.

**How is spectral flow cytometry different from cytometry with flow cytometry filters?**

It measures each event's whole emission spectrum across many detectors and solves for fluorophore contributions, rather than assigning one filtered detector per dye and compensating spillover. That allows larger panels and dyes with overlapping peaks.

**Do I still need the compensation controls flow cytometry has always used?**

More than before. Unmixing is only as good as the reference signature of each fluorophore, so every dye in the panel needs a single stained control on a carrier that matches the experiment.

**What does autofluorescence extraction do?**

It treats the sample's own fluorescence as another signature to be solved out, using an unstained control from that sample type. On tissue and fixed samples it can recover dim populations that would otherwise be buried.

**Does spectral cytometry remove the need for careful panel design?**

No. It changes the constraint from filter availability to spectral similarity. Two dyes with nearly identical full spectra remain inseparable, and spread still costs resolution on the dimmest markers.

## Cation exchange

Cation exchange separates on net positive charge, which makes it powerful and makes it fail in one particular way: if the load conductivity is too high, nothing binds, and the target appears in the flow-through while the method looks correct on paper. This page covers setting the pH, choosing the resin and designing an elution that separates rather than merely eluting.

1. **Set the loading pH from the isoelectric point**: The protein must carry net positive charge, so load roughly one to one and a half pH units below its isoelectric point. Calculated values from sequence are approximate; measuring the apparent value by isoelectric focusing or a small pH scouting run is worth the day it takes.
2. **Get the conductivity down before the load**: Ionic strength competes directly with binding. The load has to be diluted or buffer exchanged until conductivity is low enough, and this is the single commonest reason a cation exchange step does not work. Measure it rather than assuming the dilution was enough.
3. **Choose strong or weak exchanger deliberately**: A strong exchanger keeps its charge across the useful pH range, which makes behavior predictable. A weak exchanger's charge varies with pH, which gives another selectivity handle and makes the method more sensitive to buffer preparation.
4. **Elute on salt or on pH, and try both**: A salt gradient is the default and usually adequate. A rising pH gradient can give sharper separation of charge variants, at the cost of a more demanding buffer system. For closely related species, pH elution is often the one that resolves them.
5. **Scout the conditions small before committing**: A short scouting series over two or three pH values and two resins costs a day on small columns and routinely finds a condition twice as selective as the obvious one. It is far cheaper than discovering the limitation at scale.

**What pH should I load at for cation exchange chromatography?**

About one to one and a half pH units below the protein's isoelectric point, so it carries net positive charge. Verify empirically, because calculated isoelectric points and observed behavior often differ.

**Why did my protein not bind?**

Load conductivity, almost always. Salt competes for the resin's charged groups, and a sample carried over from a previous step is usually far too conductive. Dilute or exchange the buffer and measure the conductivity before loading.

**Cation or anion exchange?**

Cation where the target is positively charged at a workable pH and the contaminants are not; anion where the reverse holds. Many purification trains use both, in either order, because the two select on opposite properties.

**Strong or weak cation exchanger?**

Strong for predictability across a wide pH range, which suits most capture steps. Weak when you need extra selectivity from the resin's own charge changing with pH, usually in polishing.

## HIC

Hydrophobic interaction chromatography works backwards compared with the ion exchange steps around it: the protein binds in high salt and elutes as the salt falls. That makes it excellent at removing aggregate and awkward to place in a train, because the buffer conditions it needs are the ones the neighbouring steps do not want. This page covers designing the step and where it belongs.

1. **Choose the salt for the effect you want and the protein you have**: Salts differ in how strongly they promote binding. Ammonium sulphate is the classic choice and a strong promoter; sodium chloride is milder and needs higher concentrations. The stronger the promoter, the greater the risk of precipitating the protein on the column.
2. **Find the binding salt concentration by a scouting gradient**: Load at increasing salt in small steps and find the lowest concentration that binds. Loading at the highest concentration the protein tolerates is a common default and a common cause of aggregate formed during the step itself.
3. **Pick ligand chemistry and density together**: Butyl, phenyl and octyl ligands differ in strength, and density matters as much as identity. A high density butyl resin can bind more strongly than a low density phenyl one, so screening across a small panel is more informative than reasoning from the ligand name.
4. **Elute on descending salt, and consider an additive**: The standard elution is a decreasing salt gradient. For proteins that will not come off, a mild additive such as a low concentration of a polyol or an organic modifier releases them, and each has to be checked against product stability.
5. **Place the step where the salt is already high**: After an ammonium sulphate precipitation or a salt elution from ion exchange, the feed is already conductive, which is exactly what this step wants. Putting it before an ion exchange step instead forces a buffer exchange nobody wanted.

**Why does hydrophobic interaction chromatography use high salt?**

Salt reduces the water ordering around exposed hydrophobic patches, which makes the protein associate with the hydrophobic ligand. Reducing the salt reverses it, which is why elution runs down a descending gradient.

**What is it best at removing?**

Aggregate, and misfolded or partly unfolded species, because those expose more hydrophobic surface than the correctly folded protein and bind more strongly. It is a standard polishing step for exactly that reason.

**How do I choose a hydrophobic interaction chromatography resin: butyl, phenyl or octyl?**

Screen them rather than reason about them. Ligand density varies between products and changes the effective strength, so a small screen across chemistries and densities with your own protein answers it faster than any rule.

**Can this step harm protein stability?**

Yes. High salt plus a hydrophobic surface is a combination that can unfold and aggregate a marginal protein. Load at the lowest salt that binds, keep the residence time short, and check aggregate before and after.

**When should I use a chromatography membrane rather than a resin?**

For polishing steps that remove low-level contaminants, where speed matters and capacity does not, and wherever single-use avoids cleaning validation. For capture of an abundant target, a packed bed still holds more per unit volume.

## Downstream bioprocessing

Downstream is where yield is lost and where cost of goods is decided, and the order of the steps matters as much as the steps themselves. A train designed step by step in isolation usually needs a buffer exchange between every pair. This page walks the conventional sequence, what each step is for, and the decisions that determine whether it survives scale-up.

1. **Clarify the harvest without shearing it**: Centrifugation then depth filtration is the conventional route for mammalian culture. Cell density and viability at harvest decide the filter area needed, and a culture harvested late shears more debris into the stream and raises the burden on every step after it.
2. **Capture on affinity, and design for what it does not remove**: Affinity capture takes the product out of a complex feed in one step at high purity. It does not remove aggregate, charge variants or leached ligand, which is what the polishing steps exist for. The eluate is also low pH, which is convenient for the next step.
3. **Run viral inactivation downstream of capture, at low pH**: Holding the low pH eluate for a defined time is the standard inactivation step for enveloped viruses, and placing it here uses the pH the capture elution already produced. The hold time, pH and temperature are validated parameters, not process convenience.
4. **Polish on two orthogonal modes**: Two steps selecting on different properties, typically an ion exchange and either a hydrophobic or a mixed mode step, remove aggregate, host cell protein and DNA. Orthogonality is the point: two steps selecting on the same property remove the same impurities twice.
5. **Concentrate and exchange by tangential flow filtration**: Tangential flow filtration concentrates the product and exchanges it into the formulation buffer in one operation. Membrane cut-off, flux and the number of diavolumes are the parameters, and the practical limit is usually the viscosity the product reaches at final concentration.

**What order should downstream bioprocessing steps go in?**

Clarification, affinity capture, low pH viral inactivation, two orthogonal polishing steps, viral filtration, then tangential flow filtration into the formulation buffer. The order is chosen so each step's output suits the next one's input.

**Why is viral inactivation downstream of capture?**

Because the affinity elution is already at low pH, so the hold costs no extra buffer exchange, and because the feed is far cleaner than the harvest, which makes the step more robust and easier to validate.

**What does tangential flow filtration (TFF) do that a column cannot?**

It concentrates and exchanges buffer without binding the product, so there is no elution step and no capacity limit in the usual sense. The limits are membrane flux and the viscosity of the concentrated product.

**How many polishing steps are needed?**

Usually two, selecting on different properties. One is sometimes enough for a very clean capture and a tolerant specification; three suggests the capture step or the upstream process is doing less than it should.

**How do I size the membrane area of a tangential flow filtration system?**

From flux measured on a small device with your real feed, not from a water flux figure. Process time is area multiplied by flux, and a real feed is usually several times slower than water.

**Tangential flow filtration cassette or hollow fibre?**

Cassettes for high area in a small footprint where the product tolerates shear; hollow fibres for shear-sensitive material and feeds carrying cells or particulates, which is why perfusion uses them.

**How is a sterilising filter's integrity proved?**

By measuring the pressure at which gas passes through the wetted membrane and comparing it against the manufacturer's value correlated with organism retention. Where the filtration supports a sterility claim, the test is run after use as well as before.

**How many buffer changes does dialysis need?**

Exchange is exponential in the volume ratio, so two or three changes of a large volume remove small molecules almost completely while one change does not. Work in changes rather than in hours.

**Does cellulose tubing need preparing?**

Yes. It is supplied with glycerol and sulphur compounds and has to be rinsed, and boiled in a chelating solution for trace-metal or nucleic acid work. Skipping it contaminates the sample.

## Preclinical CRO

Preclinical work is outsourced for capacity and for models, and the commonest expensive mistake is choosing on price for a study whose result will be used to make a go or no-go decision. The model, the conduct standard and the contract each decide whether the data survives scrutiny. This page covers the three in the order they should be settled.

1. **Start from the question, and let it choose the model**: Efficacy, mechanism and safety ask different things. Disease models vary enormously in how well they predict, and a provider's familiarity with one model is worth more than a slightly lower price on another. Ask what they run weekly, not what they can run.
2. **Specify xenograft work properly or expect to repeat it**: A xenograft study's result depends on the cell line and its authentication, the host strain and its immune status, the implantation site, the randomisation point by tumour volume and the endpoint definition. Any of these left to the provider's default is a variable you cannot reconstruct later.
3. **Decide the conduct standard deliberately**: Pivotal safety studies supporting a regulatory submission are conducted under good laboratory practice. Exploratory efficacy work generally is not, and conducting it to that standard adds cost without adding value. Deciding this late means repeating studies.
4. **Read the animal welfare and quality position before the quotation**: Institutional approval, facility accreditation, veterinary oversight and the humane endpoints in the protocol are documents, not assurances. They will be asked for later by a reviewer or a regulator.
5. **Settle data, samples and intellectual property in the contract**: Who owns the raw data, who holds the residual samples and for how long, what happens to a derived model or cell line, and who may publish. These are negotiable at contract and effectively fixed afterwards.

**How do I choose a preclinical contract research organization for my disease models?**

On model experience first, quality systems second, price third. A provider running your model routinely will produce cleaner data and spot a problem sooner than one running it for the first time at a lower rate.

**Does every preclinical study need GLP?**

No. Good laboratory practice applies to the safety studies supporting a submission. Exploratory efficacy and mechanism work is usually not conducted under it, and conducting it there adds cost without improving the science.

**What should a xenograft study protocol fix in advance?**

Cell line identity and authentication, host strain, implantation site and cell number, the tumour volume at which animals are randomised, the measurement schedule, the humane endpoint and the statistical plan. Each one changes the result.

**Who owns the data from contract research organization services?**

Whatever the contract says. Assume nothing: raw data ownership, residual sample custody, the right to publish and the ownership of anything derived are separate clauses and defaults differ between providers.

**Which humanised model should I use?**

Mature peripheral cells for short studies of human T cell function, accepting the graft-versus-host window. Stem cells for a longer-lived, more complete system, accepting months to engraftment and T cells educated on mouse tissue.

## Lab water purification

Water is the most used reagent in the building and the least specified. The grades are defined, the systems that make them are well understood, and almost every problem traced back to water is either the wrong grade for the application or contamination introduced after the purifier. This page covers the grades, sizing and where the quality is actually lost.

1. **Match the grade to the application**: Type 1 water, at about eighteen megohm-centimetres resistivity with low total organic carbon, is for the sensitive work: chromatography, molecular biology, cell culture, trace analysis. Type 2 serves as feed water, for buffers and for glasswashing. Type 3 is for rinsing and baths.
2. **Size on the daily volume and on the peak**: Systems are specified by production rate and reservoir volume, and a laboratory whose whole demand arrives in one hour of the morning needs a bigger reservoir rather than a bigger purifier. Measure a week of real consumption before ordering.
3. **Establish the feed water quality first**: Mains water hardness, chlorine and silica decide the pretreatment and the consumable life. A system quoted against a national average feed and installed on hard water will consume cartridges at a rate nobody budgeted for.
4. **Put the polishing at the point of use**: Purity is lost in storage and distribution, not in the purifier. A polishing cartridge and a final filter at the tap are what make the delivered water match the specification, and a system without them delivers reservoir water.
5. **Choose the final filter for what you are protecting against**: A 0.22 micrometre filter for bacteria, an ultrafilter where nucleases or endotoxin matter, and an ultraviolet lamp where organic carbon does. Each addresses a different contaminant and none substitutes for another.

**What is the difference between a type 1 water purification system and a type 2 one?**

Type 1 delivers about eighteen megohm-centimetres with very low organic carbon for sensitive applications. Type 2 is a lower grade, typically produced by reverse osmosis with deionisation, used as feed for type 1 systems and for general laboratory work.

**Does resistivity tell me the water is pure?**

Only about ions. Resistivity says nothing about organic carbon, bacteria, endotoxin or nucleases, and water reading eighteen megohm-centimetres can still be unusable for cell culture or for RNA work.

**How is an ultra pure water purification system sized?**

By daily volume and by peak demand together. The production rate covers the day and the reservoir covers the hour when everybody draws at once, and getting the second wrong is what produces queues at the tap.

**Where do laboratory water purification systems usually go wrong?**

After the purifier. Storage, distribution loops and the tap itself reintroduce organics and bacteria, which is why point of use polishing and a final filter matter more than another stage in the cabinet.

**Do I need the highest grade everywhere?**

No. Run a general supply for washing and baths with a polisher at the point of use. Which specification matters differs by application: organics for trace analysis, endotoxin for culture, nuclease activity for molecular work.

## Glassware washers

A glassware washer is bought as a box and lives or dies on its racks. Narrow-necked vessels only get clean if water is injected into them, and glassware only stays clean if the final rinse is purified water. Everything else in the specification is secondary. This page covers what to specify, and how to demonstrate that what comes out is actually clean.

1. **Start with the glassware you actually wash**: Volumetric flasks, measuring cylinders, bottles and pipettes each need a different injection spindle. Count the items by type and specify the racks first, because a washer with the wrong racks cleans the outside of your flasks very effectively.
2. **Insist on a purified water final rinse**: Mains water leaves mineral residue as it dries, and that residue is a contaminant in trace analysis and cell culture alike. A final rinse from a type 2 or better supply is the difference between clean glassware and glassware with a film.
3. **Decide whether you need thermal disinfection**: A hot cycle holding around ninety degrees provides thermal disinfection for laboratories handling biological material. It also rules out heat sensitive plastics, so the load mix has to be settled before the cycle set is chosen.
4. **Specify drying, and check where the air comes from**: Hot air drying inside the chamber is what prevents recontamination on the bench. Check that the drying air is filtered, because an unfiltered blower puts room dust into everything that has just been cleaned.
5. **Plan the services before the delivery**: Water, drain, power and often a purified water feed have to be at the location. Three phase supply and a floor drain are the two that most often turn a straightforward installation into a building project.

**Can a domestic glass dishwasher be used for laboratory glassware?**

Not for anything that matters. It has no injection spindles for narrow necks, no purified water rinse, no validated cycles and no documentation. It washes the outside of the glass and leaves mineral residue as it dries.

**What makes laboratory glassware washers expensive?**

Racks, the purified water circuit, the documentation and cycle control. The chamber is the cheapest part, which is why comparing machines on chamber size alone is misleading.

**How do I prove the glassware is clean?**

Test rather than assume. A residue check on the final rinse water, a wipe test for the detergent, and a functional check such as running a known blank in cleaned glassware. For regulated work these become a documented cleaning validation.

**How long is a typical cycle?**

Often well over an hour once wash, rinses and drying are included, which is why throughput is a rack loading question rather than a cycle time question. Two machines running short loads usually beat one running long ones.

**How do I compare laboratory glassware washer manufacturers?**

On the racks and injector sets configured against your own glassware inventory, and on the final rinse water supply. The cabinets are largely alike; the rack decides whether your awkward items get cleaned.

## Lab refrigerators and freezers

Cold storage is where samples are lost, and almost always for one of three reasons: a domestic appliance with a cycling temperature, a unit sited where it cannot reject heat, or an alarm that nobody received. This page covers what separates laboratory grade cold storage from an appliance, and the monitoring that turns a cabinet into controlled storage.

1. **Insist on forced air circulation and a stated uniformity**: Domestic refrigerators cycle over a wide band and stratify. A laboratory unit circulates air and states its uniformity and stability, which is the specification that decides whether a sample at the back of the bottom shelf saw the temperature the label claims.
2. **Match the interior to what goes in it**: Flammable solvent storage needs a spark free interior; an explosion proof unit is a different and much more expensive category required only where the room atmosphere is hazardous. Storing flammables in a domestic refrigerator is a recognized cause of laboratory fires.
3. **Specify an access port and use it**: A port for a probe lets an independent monitor read the actual storage space rather than the unit's own control sensor. Without one, the monitoring measures whatever the manufacturer decided to measure.
4. **Decide where the heat goes before choosing the footprint**: An undercounter laboratory freezer is convenient and rejects its heat into the leg space of the bench, which matters in a small room. Check the clearance requirement and the room's cooling capacity before deciding, because a unit that cannot reject heat runs continuously and fails early.
5. **Design the alarm around who will answer it**: An alarm is worth what the response to it is worth. Specify the escalation path, the out of hours contact and the battery backup for the monitor itself, and test the whole chain rather than the buzzer.

**Why can I not use a domestic refrigerator for samples?**

Because it cycles over a wide temperature band, stratifies badly, has no uniformity specification, no alarm and no port for independent monitoring. For flammables it is also an ignition risk.

**When is an undercounter laboratory freezer the right choice?**

When the volume is genuinely small and bench-side access matters. Capacity per unit of floor space is poor and heat rejection into a small room is a real constraint, so it suits a satellite location rather than a main store.

**What temperature monitoring do laboratory freezers need?**

Continuous logging from a probe in the storage space, not the unit's own display, with alarms that reach somebody who can act. For regulated or clinical material the logged record is part of the evidence the material is fit to use.

**How much spare capacity should be planned?**

Enough to absorb the contents of the largest unit if it fails. A cold store with no failover means a failure becomes a loss, and that plan has to exist before the failure rather than during it.

**Is a laboratory deep freezer different from a low temperature cabinet?**

The names overlap and the specification does not. What matters is the temperature the cabinet holds with the door opened at the frequency you will open it, the pull down time after a door event, and whether it holds that temperature at the top of the ambient range the room reaches in summer. Ask for the recovery curve rather than the setpoint.

## Choosing a cytometer

A cytometer is chosen by the panel it has to run, and a panel designed after the instrument arrives is a panel constrained by somebody else's optical layout. This page reads four analysers on their manufacturers' own pages and says what each is built around, so the shortlist starts from the detection you need rather than from a brand.

1. **Cytek Aurora, full spectrum detection**: Cytek presents the Aurora as a full spectrum instrument, with its own published material on enhanced small particle detection and on spectral analysis applied to cytometry. Read it as the spectral-first option: the design assumption is that you will unmix whole emission signatures rather than assign one filtered detector per dye.
2. **Beckman Coulter CytoFLEX, a configurable conventional analyser**: Beckman's CytoFLEX page describes a violet-blue-red series activatable from four up to thirteen colors, listing thirteen bandpass filters by wavelength, with an optional plate loader for throughput and software tools for 21 CFR Part 11 environments. Read it as the conventional analyser you buy small and light up later.
3. **BD FACSymphony A5 SE, spectral or compensation on one instrument**: BD describes the A5 SE as offering full visible spectrum coverage without filter changes while retaining the familiar compensation workflow, unmixing spectral data live in its acquisition software, with autofluorescence unmixing offered as a way to resolve difficult populations and an upgrade kit route from an existing A5. Read it as the option for a laboratory that wants spectral capability without abandoning its established workflow.
4. **Thermo Fisher Attune, acoustic focusing for throughput**: Thermo's Attune family page leads on acoustic focusing, using sound waves rather than hydrodynamic focusing to align cells, claiming faster processing and higher sample throughput, across the NxT workhorse, the CytPix model with a high speed brightfield camera and image analysis, and the Xenith. Read it as the throughput-and-imaging option.
5. **Then decide sorting, containment and who runs it**: An analyser and a sorter are different purchases with different rooms. A sorter generates aerosols and needs containment and a trained operator, which is why many laboratories buy an analyser and buy sorting as a service from a core facility.

**What is the best flow cytometer?**

The one whose laser and detector configuration runs the panel you need, in the room you have, with the service you can get. Of the four read here, two lead on spectral detection, one on configurable conventional detection and one on acoustic focusing for throughput, and those are genuinely different jobs.

**Spectral or conventional flow cytometry equipment?**

Spectral for large panels, for dyes with overlapping peaks and where sample autofluorescence is a problem. Conventional where panels are modest and the workflow is established. One instrument read here deliberately offers both workflows on the same machine.

**How many detectors do I actually need?**

Design the panel first, then count. Buying detectors you may use later is how laboratories end up with an instrument configured for someone else's experiment, and most platforms allow channels to be activated later anyway.

**Do I need 21 CFR Part 11 tools?**

Only if the data supports regulated work. One of the four read here names Part 11 tooling in its software on the product page; for a research laboratory that capability is usually cost without benefit.

**What should the quotation include beyond the instrument?**

Installation and qualification, the service contract, the sheath and waste handling, the computer and analysis software licenses, and training. On a cytometer the service contract is a large recurring figure and it belongs in the comparison from the start.

## Clinical sample management

A clinical sample carries obligations a research aliquot does not: somebody consented to a defined use, a coding key exists somewhere, and both the sample and its record have to survive the study and then be disposed of properly. Most failures here are administrative rather than technical, and they are discovered when somebody asks whether a sample may be used for a new analysis. This page covers the record that answers that question.

1. **The consent, and what it actually permits**: Record the protocol, the approving committee and the scope of the consent against the sample, not against the study folder. The question asked later is always about one sample and one proposed use, and a system that cannot answer it forces a conservative no.
2. **Coded, not anonymised, and know the difference**: A coded sample can be linked back to the donor through a key held separately; an anonymised one cannot. That distinction decides which regulations apply and whether results can ever be returned. Record which one each sample is, and who holds the key.
3. **Custody as structured data**: Every transfer with who released it, who received it, when and at what temperature. Custody that lives on paper forms scanned into a folder cannot be queried, and it is queried exactly when a shipment arrives warm and somebody has to decide whether the material is usable.
4. **Storage location down to the position**: Freezer, rack, box, position, with the temperature history of that freezer alongside. A sample whose location is a freezer number is a sample somebody will spend forty minutes finding with the door open, and the excursion that causes is real.
5. **The end of life, planned at the start**: Consent usually has a horizon and protocols specify destruction or return. Record the disposition date at accession and act on it. Indefinite retention of human material nobody has a current permission for is the failure that surfaces in an audit.

**What does clinical sample management have to record that research sample management does not?**

The consent and its scope, the coding status and who holds the key, the chain of custody as data, and the planned disposition. Those four answer the only question that matters later: may this sample be used for this.

**Coded or anonymised?**

Coded samples can be re-linked through a separately held key and remain identifiable in regulatory terms. Anonymised samples cannot be re-linked at all, which removes obligations and also removes the ability to return a result or add a clinical variable later.

**Do we need a separate system from the research LIMS?**

Not necessarily, but the consent, custody and disposition fields have to exist and be enforced. A research system with those added as free text will not stop a sample being used outside its consent, which is the whole point.

**How long should samples be kept?**

For as long as the protocol and the consent permit, and no longer. Record the date at accession, review it on schedule, and document the destruction. Keeping material because it might be useful is not a retention policy.

## Antibodies for niche targets

For a heavily used target there are published images, community forums and a consensus clone. For a target like these there is a catalog entry, a datasheet image and very little else, and the buyer has to do the validation that the crowd usually does. This page sets out how to judge such a reagent from what is available, and what to run before the antibody is allowed to decide anything.

1. **Read the immunogen, not the marketing**: Which region of the protein, and is it shared with the family? Enzymes and regulators frequently sit in families with high similarity, so an antibody raised against a conserved domain will recognize the relatives too. If the immunogen sequence is not disclosed, that itself is information.
2. **Check the applications claimed against the evidence shown**: A datasheet listing six applications with one western blot image has evidence for one. Treat each application as separate: performance on a denatured blot says very little about performance on a fixed section or in a pulldown.
3. **Insist on a genetic negative where one is possible**: A knockout or knockdown sample is the only control that distinguishes specific binding from a band of the right size. Where no knockout exists, an overexpression sample is the next best thing and considerably better than nothing.
4. **Prefer a recombinant monoclonal for a long project**: For a target you will work on for years, a defined sequence removes lot variation entirely. A polyclonal against a niche target may be a single bleed, and when it is gone the project restarts its validation from the beginning.
5. **Buy one lot, and bank it**: Ask the supplier how much of the lot remains before committing. For a low-demand target the answer is often small, and discovering that halfway through a study is the expensive version of this conversation.

**How do I validate a CYP26B1 antibody, or any antibody nobody has published on?**

With a genetic negative if one exists, an overexpression positive if it does not, a size check against the predicted mass with the modifications you expect, and a second antibody against a different epitope. Agreement between two independent antibodies is the strongest evidence available without a knockout.

**Is a polyclonal or a recombinant monoclonal better for a niche target?**

A polyclonal is often more sensitive and is finite. A recombinant monoclonal is reproducible indefinitely and can be reformatted. For a multi-year project the recombinant is usually the cheaper decision by the end.

**What if a RAB27B antibody detects a band of the wrong size?**

Check the predicted mass against known isoforms and modifications before concluding anything, then run the genetic control. A band at an unexpected size that disappears in a knockout is real biology; one that does not is the antibody.

**Does a supplier's validation badge mean the antibody is validated?**

It means the supplier applied its own scheme, and those schemes differ a great deal. Read what was actually done and in which application, and treat it as a starting point for your own control rather than as a substitute.

## Matrix and receptor antibodies

These three targets are grouped here because they break the default protocol in three different ways: one is secreted and mostly measured in fluid, one carries so much glycan that it runs as a smear, and one is a membrane receptor whose antibodies are often conformation dependent. Knowing which problem you have before you buy saves a great deal of time.

1. **Osteoprotegerin: a secreted decoy receptor, usually measured in fluid**: It circulates and binds its ligand, so the honest measurement is usually an immunoassay on serum or conditioned medium rather than a tissue stain. A matched standard and a matrix-matched control are what make the number mean anything, and a western blot of tissue lysate answers a different question.
2. **Decorin: a proteoglycan that will not run as a band**: The core protein carries a glycosaminoglycan chain that makes it migrate as a broad smear at a much higher apparent mass than the sequence predicts. Enzymatic removal of the chain before electrophoresis converts the smear into a band, and an antibody raised against the core protein may need that treatment to see anything at all.
3. **CD36: a membrane scavenger receptor with conformational epitopes**: Many antibodies against it recognize the folded extracellular domain, so reducing and boiling the sample can abolish the signal. Test non-reducing conditions and a gentler denaturation before concluding the antibody does not work, and check whether the clone was raised for flow cytometry rather than for blotting.
4. **Match the antibody to the application, not to the target**: For all three, an antibody sold for flow cytometry, one for immunohistochemistry and one for immunoassay capture are frequently different clones with different evidence. Buy against the application you will run and read the evidence for that application specifically.
5. **Run the sample-type control, not just the reagent control**: For a secreted target, a depleted or blank matrix. For a proteoglycan, a deglycosylated and an untreated lane side by side. For a surface receptor, a cell line known not to express it. These are the controls that actually interpret the result.

**Why does my decorin antibody give a smear rather than a band?**

Because the glycosaminoglycan chain on the core protein makes migration heterogeneous. Treating the sample to remove the chain before electrophoresis collapses the smear into a band at the predicted core mass.

**Should osteoprotegerin be measured by western blot or immunoassay?**

Usually immunoassay, because it is secreted and the question is normally how much is present in a fluid. A blot tells you about a lysate, which for a secreted protein is rarely the sample you care about.

**Why did my cd36 antibody stop working after I changed the protocol?**

Most likely you started reducing or boiling the sample, and the clone recognizes a conformational epitope on the folded extracellular domain. Try non-reducing conditions and a lower denaturation temperature.

**Can one antibody serve flow cytometry and western blotting?**

Sometimes, and it has to be shown rather than assumed. A clone selected against native surface protein and one selected against a denatured fragment are answering different questions, and the datasheet should show evidence for each application separately.

**Does a binding protein measurement tell me about the hormone?**

Only indirectly. Binding proteins report carriage and availability, and their levels move with nutrition, inflammation and liver function on their own. Where the claim concerns signalling, measure the partner as well and say which was measured.

## Therapeutic target antibodies

Both of these targets have approved therapeutic antibodies against them, which means the catalog is full of reagents whose descriptions borrow therapeutic language. A detection antibody and a function-blocking antibody are different products with different evidence, and buying the first when you needed the second is the most common mistake on targets like these.

1. **Decide whether you need detection or neutralisation**: A detection antibody binds the target on a blot, a section or in an assay. A neutralising antibody blocks a function, and that has to be demonstrated in a functional assay with a dose response. A datasheet that does not show a neutralisation curve has not shown neutralisation.
2. **Check which isoforms and processed forms are covered**: Both targets exist in multiple forms: one is proteolytically processed, the other has several splice variants with different receptor binding. An antibody against one region may see only some of them, and the region matters more than the target name.
3. **Read the species claim as tested or predicted**: For work in a model organism, reactivity has to be tested rather than inferred from sequence similarity. Ask for the data in that species, particularly for neutralisation, where potency frequently differs even when binding does not.
4. **Demand a low endotoxin, carrier-free preparation for cell work**: Any antibody added to living cells needs a stated endotoxin limit and no bovine carrier protein. Carrier protein also blocks conjugation chemistry later, so specify carrier-free if there is any chance of labeling it.
5. **Pair it with an isotype control from the same supplier**: An isotype and species matched control at the same concentration, ideally the same format and endotoxin grade, is what separates a functional effect from an immunoglobulin effect. Buy it at the same time or it will not be run.

**Is a pcsk9 antibody the same as the therapeutic?**

No. Research reagents against this target are for detection or for blocking in an experiment; the approved therapeutics are separately engineered, humanised and manufactured molecules. Do not read clinical evidence as evidence about a catalog reagent.

**How do I know a vegf antibody actually neutralises?**

A functional assay with a dose response and a stated half-maximal inhibitory concentration in a named system. Binding data, however good, does not demonstrate blocking, and the two are frequently presented side by side as if they did.

**Does it matter which isoform the antibody sees?**

It usually decides the experiment. Both of these targets have forms with different biology, so an antibody's epitope region tells you more about what it will report than the target name does.

**What grade do I need for adding to cells?**

Carrier-free, with a stated endotoxin limit and sterile filtration, and azide-free. Azide is toxic to cells and is present in many general-purpose preparations.

## Human ELISA kits

Most ELISA disappointments are range and matrix problems rather than kit quality. A target present at grams per litre and a target present at picograms per millilitre need completely different kits, and a kit validated in serum may behave differently in culture medium or in lysate. This page covers choosing the kit and proving it works on your own samples.

1. **Start from the expected concentration, not the target name**: Albumin in serum is one of the most abundant proteins there is and needs enormous dilution; a neurotrophin or a cytokine sits at the other end of the scale. Pick the kit whose standard curve brackets your expected values in the middle of its range, where precision is best.
2. **Check the matrix the kit was validated in**: Serum, plasma with a named anticoagulant, culture supernatant, lysate and urine are different matrices and each can interfere differently. A kit validated in serum used on culture medium is an unvalidated assay until you show otherwise.
3. **Run spike recovery and linearity of dilution on your own samples**: Spike a known amount into your matrix and recover it, and dilute a high sample serially and check the back-calculated values agree. Recovery within roughly eighty to a hundred and twenty percent and parallel dilution are the two tests that qualify a kit for your sample type.
4. **Decide what the kit actually detects**: Free against total, mature against precursor, and whether a binding protein in the sample masks the target. For several cytokines and neurotrophins this single question explains most of the disagreement between published values.
5. **Fix the handling before the first plate**: Tube type, time to centrifuge, storage temperature and the number of freeze-thaw cycles all move results, and several cytokines are released from cells during clotting. Write the pre-analytical protocol down and follow it for every sample in the study.

**How do I choose between two human elisa kits for the same target?**

Compare the standard range against your expected concentration, the validated matrices, the published spike recovery and dilution linearity, and what the antibody pair detects. Price rarely separates them usefully; range and matrix almost always do.

**Why do two BDNF ELISA kit results differ for the same sample?**

Different antibody pairs see different forms, and different standards are calibrated differently. Values from two kits are generally not interchangeable, which is why a study should use one kit and one lot throughout.

**What does spike recovery tell me in ELISA assay validation?**

Whether your matrix interferes. Adding a known amount of standard to your sample and recovering roughly what you added shows the matrix is not suppressing or enhancing the signal; a recovery well outside that says it is.

**Can I use a human interferon gamma ELISA kit validated in serum on culture supernatant?**

Only after qualifying it. Medium contains serum components, phenol red and additives that can interfere, so run spike recovery and dilution linearity in your actual medium before trusting the numbers.

## Bulk RNA sequencing

Bulk RNA sequencing is cheap enough that the design decisions now cost more than the sequencing. The three that matter are how the library is made, how deep it is sequenced and how many biological replicates there are, and only the last of those is usually short. This page covers the choices in the order they have to be made.

1. **Choose the library chemistry from the RNA you have**: Poly-A selection with an mrna isolation kit captures intact polyadenylated transcripts and is the efficient choice for good quality eukaryotic RNA. A ribosomal rna depletion kit keeps non-polyadenylated and degraded material, which is what degraded or archival samples, bacterial RNA and non-coding work require.
2. **Measure RNA integrity before you commit**: An integrity number from a capillary system, not a gel photograph. Degraded RNA is where poly-A selection fails quietly, giving a strong three prime bias that looks like biology in the coverage plot and is not.
3. **Set depth from the question**: Differential expression of moderately expressed genes is generally served by something in the region of twenty to thirty million reads per sample. Isoform level work, low abundance transcripts and allele specific analysis need considerably more, and going deeper never substitutes for having too few samples.
4. **Spend the marginal dollar on replicates**: Below about four or five biological replicates per group, power is dominated by biological variance rather than by counting noise, so an extra sample buys far more than an extra ten million reads. This is the most common and most expensive design mistake in the field.
5. **Decide stranded and read length deliberately**: A stranded protocol should be the default: it resolves overlapping genes and antisense transcription for no extra cost. Paired end and longer reads help isoform assignment and are largely wasted on straightforward gene level counting.

**Should I use an mRNA isolation kit or a ribosomal RNA depletion kit?**

Poly-A with an mRNA isolation kit for intact eukaryotic RNA and gene level expression. rRNA depletion for degraded or archival samples, for bacterial RNA, and whenever non-polyadenylated transcripts are part of the question.

**How many reads per sample do I need?**

For differential expression of moderately expressed genes, roughly twenty to thirty million is a common working figure. Isoform level and low abundance work need more. Decide from the question and then check what the analysis actually requires.

**How many replicates?**

More than three if the effect is modest. Biological variance, not sequencing depth, sets the power in almost every bulk experiment, so the marginal sample is worth far more than the marginal read.

**Does batch matter if everything is sequenced together?**

Library preparation batch matters as much as sequencing batch, and so does the day of extraction. Randomise groups across every batch boundary and record them, so batch can be modelled rather than confounded.

## Automated extraction

Automating extraction is bought for consistency as much as for hands-off time: a bead protocol run identically every day removes the operator variance that shows up downstream as unexplained inhibition and variable yield. This page covers when it pays, which chemistry suits which sample, and the questions that separate platforms once the brochures are set aside.

1. **Count the real daily demand, and the shape of it**: A platform that processes twenty four samples per run suits a laboratory with a steady flow; one that processes ninety six suits batching. A laboratory whose work arrives as four samples at a time all day will run an expensive instrument at a fraction of its capacity.
2. **Choose the chemistry from the sample, not the instrument**: Magnetic bead chemistry dominates automation because it needs no centrifugation and scales cleanly. Column chemistry gives excellent purity for some difficult samples. Match it to your hardest sample type, because that is the one that will fail.
3. **Check the validated protocols you actually need**: Blood, tissue, swabs, saliva, plant material and stool all behave differently and each needs its own validated protocol. A platform with one generic protocol will need method development for everything else, and that is the hidden cost.
4. **Take contamination control seriously on a shared deck**: An open deck processing amplifiable material next to extraction is a cross contamination risk. Closed cartridges remove most of it at a higher consumable cost. Decide which matters more before the platform is chosen rather than after the first false positive.
5. **Price the consumables over five years**: Cartridges and plates are where the money goes, and they are proprietary on most platforms. A cheap instrument with expensive locked-in consumables is frequently the costlier choice, and the sum is easy to do before purchase.

**When does an automated nucleic acid extraction system pay for itself?**

Usually at a few dozen samples a day, and sooner where consistency rather than hands-off time is the driver. Below that, the consumable premium and the maintenance rarely beat a good manual protocol and a careful operator.

**Magnetic beads or columns?**

Beads for almost all automation, because they need no centrifugation and scale cleanly across deck formats. Columns still win on some difficult matrices where their purity is genuinely better, and a platform that supports both keeps that option open.

**Does automation improve yield?**

Not usually. It improves consistency, which is often the more valuable property: a tighter distribution of yield and purity makes downstream quantification and normalisation far more reliable, even when the mean is unchanged.

**What about cross contamination?**

It is the real risk on an open deck, particularly in a laboratory that also amplifies. Closed cartridge systems, physical separation of pre and post amplification areas, and regular environmental swabbing are the controls that matter.

## Automated western blot

Automated western blotting replaces the gel, the transfer and the membrane with a capillary in which separation and immunodetection both happen, and reports a quantified peak rather than an image of a band. The reproducibility gain is genuine and so are the constraints, and both are usually understated by whoever is selling the instrument. This page covers what changes.

1. **Understand what the output actually is**: A capillary system reports a signal against apparent molecular weight as a trace, with the band image reconstructed for familiarity. That makes quantification straightforward and makes some judgements a person makes by eye on a membrane, such as recognising a smear, harder.
2. **Check your antibodies transfer to the format**: Not every antibody that works on a membrane works in a capillary. The chemistry and the matrix differ, and each antibody has to be re-titrated and sometimes replaced. Budget that revalidation into the purchase rather than assuming the panel carries over.
3. **Match it to small, precious samples**: The strongest case is very small input: a few microlitres of lysate per run makes the format attractive for rare cells, biopsies and microdissected material where a conventional blot is simply not possible.
4. **Be realistic about throughput and cost per sample**: A run processes a fixed number of capillaries in a few hours and the consumable cost per sample is substantially higher than a gel. For a laboratory running many samples of one abundant target, a gel remains cheaper.
5. **Keep the loading control question in view**: The same normalisation argument applies here as on a membrane: a housekeeping protein has to be genuinely invariant, and total protein normalisation is available on these systems and is frequently the better answer.

**What does an automated western blot system do differently?**

It separates proteins by size in a capillary and performs immunodetection inside that capillary, reporting a quantified trace. There is no gel, no transfer and no membrane, which removes the two steps that cause most of the variability in a conventional blot.

**Will my antibodies work on it?**

Some will and some will not, and it has to be tested. Expect to re-titrate every antibody and to replace a proportion of them, and treat that as part of the setup cost.

**Is this western blot alternative more sensitive than a conventional blot?**

Often, particularly for small sample amounts, but the real gain is reproducibility and quantification rather than raw sensitivity. Claims of large sensitivity gains should be tested on your own lysates.

**Should it replace conventional blotting, or a western blot service?**

Rarely. Most laboratories that adopt it keep gels for screening, for large sample numbers and for anything where seeing the whole lane matters, and use the automated system where quantification or sample scarcity drives the work.

## RNase inhibitors

Two enzymes are added to almost every RNA workflow for opposite reasons, and both have failure modes that present as a failed experiment rather than as a reagent problem. The inhibitor stops working silently under conditions people routinely use, and the nuclease that removes contaminating DNA will remove your RNA if it is inactivated the wrong way. This page covers both.

1. **Know which inhibitor class you are buying**: Protein ribonuclease inhibitors bind the enzyme tightly and are effective and fragile: they lose activity below about pH seven, above roughly fifty degrees, and in the presence of reducing-agent-free buffers, because they depend on free thiols. Chemical inhibitors are more robust and less specific.
2. **Keep the reducing agent in the buffer**: A protein inhibitor oxidises and dissociates without a reducing agent present, at which point it releases the ribonuclease it was holding. A buffer prepared without one is worse than no inhibitor at all, because the protection was assumed.
3. **Do not heat it and expect it to work**: Reverse transcription at the higher temperatures some enzymes allow is outside the protein inhibitor's working range. Where the protocol runs hot, the inhibitor is protecting the sample before the reaction, not during it.
4. **Use dnase i to remove genomic DNA, deliberately**: Residual genomic DNA is the usual cause of a signal in a no-reverse-transcription control. A DNase step during or after purification removes it, and a no-RT control on every experiment is what proves the step worked.
5. **Remove the nuclease without boiling the RNA**: Heating DNase I in the presence of magnesium degrades RNA. Inactivate by chelating the magnesium, by a column or bead cleanup, or with a removal reagent, and treat a 95 degree kill step as the thing that ruined the sample rather than as a protocol.

**Which rnase inhibitor should I use?**

A protein inhibitor for standard reactions near neutral pH and moderate temperature, with a reducing agent in the buffer. A chemical inhibitor where conditions are harsher or where you cannot guarantee the buffer, accepting that it is less specific.

**Why did my RNA degrade despite adding an inhibitor?**

Most often the buffer had no reducing agent, the pH was below the inhibitor's working range, or the reaction ran hot. All three release the inhibitor from the ribonuclease it was holding, which is worse than never adding it.

**Do I always need dnase i?**

Whenever the assay cannot distinguish RNA from DNA, which includes most quantitative PCR against genes with no intron between the primers. Run a no-reverse-transcription control and let that decide rather than assuming.

**How should DNase be inactivated before reverse transcription?**

By removing the magnesium or the enzyme, not by heating. Heat plus magnesium fragments RNA, and this single step accounts for a great deal of unexplained loss of integrity between extraction and reverse transcription.

## Immunofluorescence

An immunofluorescence image is persuasive whatever it shows, which is exactly why the controls matter more here than in almost any other assay. Most of the decisions that determine whether the image means anything are made before the primary antibody goes on: how the sample was fixed, how it was permeabilised, and what the secondary could bind. This page covers those in order.

1. **Choose fixation for the epitope and the structure**: Aldehyde fixation preserves structure by cross-linking and can mask epitopes, which is what retrieval exists to reverse. Alcohol fixation precipitates protein, preserves some epitopes better and disturbs membranes and cytoskeleton. Test both on a known positive rather than inheriting a protocol.
2. **Permeabilise according to where the target is**: A surface antigen may need no permeabilisation at all, and permeabilising it can move the signal. A nuclear target needs a stronger detergent than a cytoplasmic one. Match the agent and the time to the compartment.
3. **Block against the actual source of background**: Serum from the species the secondary was raised in blocks non-specific binding; a protein block handles general stickiness; and for tissue with endogenous biotin or strong autofluorescence, neither helps and a specific treatment is required.
4. **Choose secondaries that cannot collide**: Cross-adsorbed secondaries raised in one host against different primary species are the clean way to multiplex. Where two primaries share a host, sequential detection with a blocking step in between is the alternative, and it needs its own control.
5. **Run the controls on the same slide and the same settings**: A no-primary control, a known positive and, where possible, a genetic negative. Acquire all of them at the settings used for the experiment, because a control imaged at a lower exposure proves nothing at all.

**Which fixation should I start with for immunofluorescence microscopy?**

Try paraformaldehyde and a cold alcohol fixation side by side on a known positive sample. Which one works is a property of the epitope, and inheriting another laboratory's choice is the commonest reason a published antibody does not work for you.

**How do I choose an immunofluorescence secondary antibody?**

Raised against the host of your primary, not raised in your sample's species, cross-adsorbed where species could collide, and conjugated to a fluorophore that suits your filters and the target's abundance. Titrate it like a primary.

**What causes high background with an immunofluorescence antibody?**

An over-concentrated secondary, inadequate blocking, endogenous autofluorescence in the tissue, or drying of the sample at any stage. Run the secondary-only control first: it separates the secondary from everything else in one experiment.

**Can I quantify an immunofluorescence image?**

Only with fixed acquisition settings, a control on every slide and a pre-declared analysis. Auto-exposure across a cohort destroys quantification, and thresholds chosen after seeing the images are not measurements.

## UV spectrophotometers

A spectrophotometer is the instrument most laboratories use most often and specify least carefully. The decisions that matter are beam configuration, wavelength range, path length and how small a volume you need to measure, and each of them changes what the reading is worth. This page covers those, and clears up a naming confusion that causes real purchasing mistakes.

1. **Decide single beam or double beam**: A single beam instrument measures blank and sample sequentially and is cheaper and perfectly adequate for stable, quick readings. A double beam compares against a reference continuously and is what you want for scans, for kinetics and for anything where lamp drift over minutes would matter.
2. **Set the wavelength range from the chemistry**: Nucleic acid and protein work lives around two hundred and sixty and two hundred and eighty nanometres, so an instrument covering roughly one hundred and ninety to eleven hundred nanometres covers ultraviolet and visible work. A visible-only instrument is cheaper and cannot do the ultraviolet measurements a molecular biology laboratory makes daily.
3. **Choose the sample format honestly**: A cuvette instrument needs a millilitre or so unless you buy microcuvettes. A microvolume instrument measures a droplet on a pedestal with a very short path length, which is transformative for precious samples and is less precise at low concentration than a cuvette.
4. **Understand the path length you are actually using**: Absorbance scales with path length, and a microvolume pedestal uses a fraction of a millimetre rather than a centimetre. Instruments correct for this in software, and comparing raw numbers between formats without accounting for it is a recurring error.
5. **Treat purity ratios as hints, not tests**: The ratio of absorbance at two hundred and sixty to two hundred and eighty nanometres indicates protein contamination and the two hundred and thirty ratio indicates organic carryover. Neither detects a co-purifying inhibitor, and a preparation with perfect ratios can still fail downstream.

**Spectrophotometer vs spectrometer: what is the difference?**

A spectrophotometer measures how much light a sample absorbs or transmits at chosen wavelengths, giving absorbance. Spectrometer is the broader word for any instrument that separates and measures a spectrum, including mass spectrometers, which measure something else entirely.

**Do I need a double beam uv spectrophotometer?**

For fixed wavelength readings on stable samples, usually not. For wavelength scans, enzyme kinetics and long measurements where lamp output drifts, yes, and the difference in data quality is obvious rather than marginal.

**Is a microvolume spectrophotometer as accurate as a cuvette?**

For concentrated samples it is close and far more convenient. At low concentration the short path length costs precision, and surface tension and evaporation on the pedestal add variability. For dilute samples a cuvette remains the better measurement.

**What does an A260 over A280 of 1.8 prove?**

That there is not much protein contamination. It says nothing about salt, phenol, ethanol or the inhibitors that cause downstream failures, which is why the ratio is a screening check rather than a release test.

**Do I need a double beam UV visible spectrophotometer for scans?**

For scanning spectra, kinetic measurements and regulated work, yes: the reference corrects drift continuously. For single-wavelength absorbance readings a single beam instrument is adequate and cheaper.

**Is a microvolume reading as good as a cuvette reading?**

For concentrated samples it is convenient and adequate. Its very short path makes it insensitive at low concentration and sensitive to drop placement and surface cleanliness, so accurate quantitative work still belongs in a cuvette.

## Stability testing

Stability testing produces one number that matters, the shelf life, and it produces it slowly. Everything in the design exists to make that number defensible: the storage conditions, the time points, the analytical methods and the evidence that the chamber held what it claimed. This page covers the design and the two things that most often invalidate a study after it has run.

1. **Choose the conditions from the climatic zone and the storage claim**: The conventional long term condition for a room temperature product is twenty five degrees at sixty percent relative humidity, with an accelerated condition at forty degrees and seventy five percent. An intermediate condition is added where the accelerated data shows significant change. Refrigerated and frozen products have their own sets.
2. **Use a stability-indicating method, and prove it is one**: The analytical method has to separate the product from its degradation products, which is demonstrated by forced degradation: acid, base, oxidation, heat and light, showing that the assay still measures the intact product. A method that cannot do this cannot support a shelf life.
3. **Fix the time points and the batches before starting**: Long term studies conventionally test at three month intervals in the first year, six monthly in the second and annually after. Three primary batches is the usual expectation for a registration study. Adding a batch later does not retrofit the design.
4. **Test in the container you will actually sell**: Stability is a property of the product in its container closure system, so a study in a laboratory bottle says nothing about a blister or a prefilled syringe. Orientation matters too where the product contacts the closure.
5. **Prove the chamber before you trust the study**: A stability test chamber needs mapping, with distributed sensors, loaded, showing that every shelf held the condition. It also needs an alarm and a documented excursion procedure, because the question asked later is always what happened during the power cut.

**What conditions does stability testing use?**

Commonly twenty five degrees at sixty percent relative humidity for long term and forty degrees at seventy five percent for accelerated, for a room temperature product, with an intermediate condition added when accelerated data shows significant change. Refrigerated and frozen products use their own sets.

**What makes a method stability-indicating?**

Demonstrated separation of the product from its degradation products, shown by forced degradation under acid, base, oxidation, heat and light with mass balance. Without that evidence the assay may be measuring degradation products as product.

**How much does the stability test chamber matter?**

Entirely. An unmapped chamber means you do not know what any sample experienced, and an excursion without a documented assessment leaves every subsequent time point open to challenge. Mapping and alarm handling are the study's foundation.

**Can accelerated stability testing alone support shelf life testing?**

It can support a provisional claim in some frameworks, and it is not a substitute for real time data. Extrapolating beyond what the guidance allows is the commonest way a shelf life claim is rejected.

## Mammalian cell culture

Almost every irreproducible cell culture result comes from one of three places: the line was not what it was believed to be, it was contaminated with mycoplasma, or it had drifted at a passage number nobody recorded. None of these is difficult to prevent and all three are common. This page covers the practices that make a culture trustworthy and the media decision that sits underneath them.

1. **Authenticate the line, then authenticate it again**: Short tandem repeat profiling identifies a human line against a reference database and is inexpensive. Do it when the line arrives, before any publication and after any incident, because misidentified and cross-contaminated lines remain a substantial fraction of what circulates between laboratories.
2. **Test for mycoplasma on a schedule, not on suspicion**: Mycoplasma does not make the medium cloudy and changes gene expression, growth and drug response profoundly. Test at least quarterly and on every new arrival, quarantine incoming lines until they pass, and treat a positive as a reason to discard rather than to treat.
3. **Bank in two tiers and work from the second**: A master bank frozen at low passage and a working bank expanded from it. Everything for the study comes from the working bank, and when it is exhausted a new one is made from the master. Working continuously from a thawed vial guarantees drift nobody can measure.
4. **Decide serum against a defined medium deliberately**: Serum is undefined, varies between lots and carries supply risk; it also supports a great many lines that a defined medium does not. If you use it, test and reserve a lot. If you can move to a defined or serum-free formulation, the reproducibility gain is real and the adaptation takes weeks.
5. **Record passage number with every experiment**: Cells change with passage, and results from passage eight and passage forty are not comparable. Recording it costs nothing and is the single most useful number in a cell culture record.

**How often should mammalian cell culture be tested for mycoplasma?**

At least quarterly for lines in continuous use, and on arrival for every new line before it enters the main incubator. Infection is invisible without a test and changes results in ways that are easy to publish and hard to retract.

**Do I really need to authenticate a line I got from a colleague?**

Especially then. Lines passed between laboratories are the ones most likely to be misidentified, and a short tandem repeat profile is cheap compared with a study built on the wrong cells.

**Serum or serum-free medium?**

Serum-free where the line supports it, for reproducibility, defined composition and supply security. Serum where the line needs it, in which case test candidate lots against your own assay and reserve the one that works.

**How many passages is too many?**

It depends on the line, and the way to know is to measure: growth rate, morphology and the assay you care about, at intervals. Set a maximum passage from that evidence and return to the working bank when it is reached.

**When is a mammalian host necessary?**

When the product needs modifications a bacterial system cannot add, and those modifications affect folding, activity, stability or immunogenicity. If they do not, a simpler host is faster and cheaper.

**Fed batch or perfusion cell culture?**

Fed batch unless the product is unstable in the vessel or the density has to be sustained. Perfusion costs more medium, needs a cell retention device and is harder to control, so it is chosen for the product rather than for throughput.

## Flash chromatography

Flash chromatography is the workhorse purification of a synthetic laboratory, and most poor separations are decided before the pump starts: too much material on too small a cartridge, a gradient guessed rather than translated from a plate, and a liquid load of something that will not dissolve. This page covers all three and the automation that removes the tedium.

1. **Run the thin layer plate first, and read it properly**: Find a solvent system putting the target at a retention factor near zero point three with the nearest impurity clearly separated. That plate is the method: a poor separation on a plate will not become a good one on a column, it will become a slow one.
2. **Size the cartridge from how hard the separation is**: For an easy separation, silica of roughly twenty times the sample mass is generous; for a difficult one, fifty times or more may be needed. Sizing on sample mass alone, ignoring how close the spots are, is the commonest cause of overlapping fractions.
3. **Translate the plate into a gradient, do not guess it**: A shallow gradient around the solvent strength that gave the right retention factor is the starting point, with a hold at the beginning to let the sample settle onto the bed. A steep gradient to save time is how two compounds that separated on a plate arrive together.
4. **Dry load anything that will not dissolve**: Liquid loading in a strong solvent pushes the band down the column before separation begins. Adsorbing the crude onto a small amount of silica or an inert medium and loading it dry gives a tight band and is worth the extra ten minutes for anything awkward.
5. **Collect on the detector you actually have**: Ultraviolet detection at the right wavelength collects most compounds; a chromophore-free product needs an evaporative detector or fraction-by-fraction plate checking. Knowing which case you are in before the run avoids collecting a hundred empty tubes.

**How big a flash chromatography cartridge do I need?**

Sized to the difficulty of the separation rather than to the mass alone. Twenty times the sample mass in silica is a starting point for an easy separation, and fifty or more for close-running compounds.

**How do I convert a TLC result into a gradient?**

Find the solvent composition giving a retention factor near zero point three, then run a shallow gradient centred on that strength, with an initial hold. The plate sets the strength; the gradient's shape sets the resolution.

**When should I dry load?**

Whenever the crude will not dissolve in a weak solvent, or when the only solvent that dissolves it is strong enough to move the band. Dry loading gives a narrow starting band and fixes most disappointing separations on its own.

**Is reversed phase worth it on a flash system?**

For polar compounds that streak on silica, yes, and most modern systems support it. The cartridges cost more and the solvent is aqueous, so the drying step afterwards is longer, which is the real trade.

## Automated purification

A chromatography system removes the tedium from purification and adds a new risk: it will execute a bad method perfectly, overnight, and present you with fractions in the morning. This page covers where automation pays in a protein expression and purification workflow, how to build a multi-step method that can run unattended, and what to monitor so a failure is caught rather than collected.

1. **Automate the steps that are identical every time**: Affinity capture on a tagged construct, buffer exchange and a polishing size exclusion step are the same operations on every batch, and that is exactly where automation pays. A step requiring judgement on each run is not the one to leave alone at midnight.
2. **Chain capture into polishing through a loop, not a person**: Most systems can collect an affinity eluate into a loop and inject it onto a size exclusion column automatically. That removes the overnight gap where an eluate sits at low pH and aggregates, which is the commonest cause of losing a protein between two good steps.
3. **Monitor more than absorbance at 280 nm**: Add conductivity and pH to the trace so the run tells you whether the gradient was delivered and whether a buffer line ran dry. A purification that failed because a bottle emptied looks like a protein problem on absorbance alone.
4. **Write pooling rules rather than pooling by eye**: Define the peak collection on absorbance thresholds and on peak width in advance. Automated pooling by a rule is reproducible between batches; pooling by whoever is in the laboratory in the morning is not.
5. **Put the safety stops in before the first unattended run**: Pressure limits appropriate to the column, an air sensor on the inlet, and a defined behavior on alarm. A system that pauses on a pressure alarm saves a column; one that pushes through destroys it and the sample.

**What does automated protein purification actually improve?**

Reproducibility between batches, and recovery on steps where time matters. The largest single gain is usually chaining capture into polishing so an acidic eluate is not left standing, which is where a good deal of protein is lost.

**Can a whole recombinant protein expression and purification workflow be automated?**

The purification can be, largely. Expression, harvest and lysis are not, and the quality of what enters the first column still decides what comes out of the last one. Automation improves consistency, it does not repair a poor feed.

**Do I need a cold cabinet for the system?**

For anything marginally stable, yes. Running a system in a cold room or in a refrigerated cabinet buys stability on long unattended runs, and many proteins that fail overnight at room temperature are fine cold.

**How do I stop an unattended run destroying a column?**

Set the pressure limit to the column's specification rather than the pump's, fit an air sensor, define the alarm behavior as pause rather than continue, and filter everything that goes on. Most overnight column deaths are a blocked frit or an emptied bottle.

## Lipid nanoparticles

A lipid nanoparticle is four lipids, a nucleic acid and a mixing process, and almost all of its behavior is decided by the ionisable lipid and by how the particle was made. This page covers what each component contributes, the process parameters that matter, and the measurements that tell you whether you have a particle worth testing.

1. **The ionisable lipid does the work**: It is neutral at physiological pH and becomes positively charged in the acidic mixing buffer, which is how the nucleic acid is captured, and again in the endosome, which is how it escapes. Its structure dominates potency and tolerability, and it is the component around which the rest is designed.
2. **The helper phospholipid and cholesterol build the structure**: The phospholipid supports the bilayer and contributes to fusion with the endosomal membrane; cholesterol fills the structure and provides stability. Their ratios are formulation variables rather than fixed constants, and they move both stability and potency.
3. **The PEG lipid sets the size and the circulation behavior**: A small molar percentage of a polyethylene-glycol lipid controls particle size during formation and limits aggregation. More of it gives smaller particles and less uptake; the amount is one of the most sensitive parameters in the formulation.
4. **Fix the nitrogen to phosphate ratio and the mixing**: The ratio of ionisable lipid nitrogen to nucleic acid phosphate determines encapsulation and charge. Microfluidic mixing with controlled flow rate ratio and total flow rate is what makes particles reproducible; bulk mixing by hand is not a process, it is an event.
5. **Measure size, dispersity, encapsulation and charge, every batch**: Hydrodynamic size and polydispersity by light scattering, encapsulation efficiency by a dye exclusion assay, and surface charge near neutral at physiological pH. A batch outside any of these is not comparable with the one before it.

**What are the components of lipid nanoparticles for mrna delivery?**

An ionisable lipid, a helper phospholipid, cholesterol and a PEG lipid, formulated with the RNA. The ionisable lipid captures the RNA at low pH and drives endosomal escape, and it is the component that most determines performance.

**What does the N to P ratio change?**

Encapsulation efficiency and particle charge, and through them potency and tolerability. It is one of the first parameters screened in any formulation development and it interacts with the mixing conditions rather than acting independently.

**Why does mixing method matter so much?**

Because the particle forms in milliseconds as the ethanol and aqueous streams meet. Microfluidic mixing controls that moment reproducibly; hand mixing does not, and batches made that way differ in size and encapsulation in ways no analysis can correct.

**What should be measured on every batch?**

Particle size and polydispersity, encapsulation efficiency, RNA integrity after formulation and surface charge. RNA integrity is the one most often skipped and the one that most often explains a batch that looks right and does nothing.

## Clinical data management

Data management is judged on one day, the day of database lock, and everything that makes that day easy was decided before the first patient was enrolled. Most of the pain in a trial's data comes from a case report form that asked an ambiguous question and an edit check nobody wrote. This page covers the decisions in the order they have to be made.

1. **Write the data management plan before enrollment**: It names the systems, the roles, the coding dictionaries and their versions, the edit check specification, the query workflow and the lock criteria. Written after enrollment starts, it documents what happened rather than governing it.
2. **Design the case report form to prevent queries, not to collect them**: Every ambiguous field becomes a query, and every query costs a site coordinator and a monitor time. Constrained lists instead of free text, one question per field, and a pilot with real site staff before go-live remove most of the query volume before it exists.
3. **Build edit checks that fire at entry**: A range check, a consistency check between related fields and a required field rule catch the error in front of the person who knows the answer. A check that runs in a nightly batch produces a query a week later, to somebody who has to find the source document again.
4. **Code adverse events and medications with a stated dictionary version**: Coding to a named version, with a documented process for terms the dictionary does not carry, is what makes the safety analysis defensible. Upgrading a dictionary mid-study is a decision with consequences and needs a documented plan.
5. **Produce submission-ready datasets in the standard from the start**: Mapping to the standard tabulation model and the analysis datasets built on it is far cheaper designed in than retrofitted. A study that collects data in its own shape and maps at the end pays for that twice, in effort and in review questions.

**What does a data management plan have to cover?**

Systems and their validation, roles and access, the case report form and edit check specification, coding dictionaries and versions, the query workflow and timelines, external data handling, and the criteria for database lock.

**How do I reduce query volume?**

At form design. Constrained fields rather than free text, one question per field, edit checks that fire at entry, and a pilot with the people who will actually fill it in. Queries are a symptom of a form that asked badly.

**Why do submission standards matter before analysis?**

Because retrofitting them is expensive and error-prone. Collecting and mapping into the standard tabulation model as the study runs means the analysis datasets and the submission package are a continuation rather than a project.

**What is required to lock a database?**

All data entered, all queries closed, all external data reconciled, coding complete and quality-reviewed, and the audit trail intact. The criteria belong in the plan so lock is a checklist rather than a negotiation.

## Materials testing

Materials work is bought by people who are usually not materials specialists, and the two failures are predictable: a test ordered without naming the standard, and a failed component cleaned up before anybody looked at it. This page covers how to specify the tests, what metallurgical work adds, and how a failure analysis should be sequenced.

1. **Name the standard, the specimen and the condition**: A tensile result is meaningless without the standard, the specimen geometry, the orientation relative to the material's processing direction, the strain rate and the temperature. A quotation for a tensile test without those is a quotation for a number, not a measurement.
2. **Check the accreditation scope covers the actual test**: Accreditation is granted for specific tests on specific materials over specific ranges. A laboratory accredited for tensile testing of steel is not thereby accredited for a polymer, and the scope document says which is which.
3. **Use metallurgical testing when the question is why, not how much**: Mechanical testing tells you what a material does. Sectioning, mounting, etching and examining the microstructure tells you why: grain size, phase distribution, heat treatment condition, inclusions and weld quality. A hardness traverse across a weld answers questions no tensile bar can.
4. **Preserve the evidence before any failure analysis begins**: Photograph the assembly in place, do not clean the fracture surface, do not fit the pieces back together, and keep any fluids or debris. The fracture surface carries the history of the failure and wiping it removes the answer. This is the single most damaging thing done to failed parts.
5. **Sequence the analysis from non-destructive to destructive**: Visual and photographic, then fracture surface examination, then chemistry, then sectioning and microstructure, then mechanical tests on the remaining material. Once the part is cut, the earlier observations cannot be repeated.

**What do I need to give a mechanical testing laboratory to order material testing?**

The test standard, the material and its condition, the specimen geometry and orientation, the number of replicates, the acceptance criteria if any, and the environment the test should run in. Without these the result cannot be compared with anything.

**What does a metallurgical testing lab add?**

Microstructure, which explains behavior. Grain size, phase distribution, heat treatment condition, inclusions and weld integrity are what tell you why a material performed as it did rather than simply how it performed.

**What should I do with a failed part before sending it to a failure analysis laboratory?**

As little as possible. Photograph it in place, do not clean or wipe the fracture surface, do not mate the fracture faces together, preserve any fluids or debris, and record who handled it. Cleaning destroys the evidence the analysis depends on.

**Does a universal material testing machine give the same result everywhere?**

Only under the same standard, geometry, rate and grip arrangement. Differences between laboratories nearly always come from specimen preparation and gripping rather than from the machine, which is why the standard has to be named in the order.

## Sterility testing

A sterility test is a small test with a large consequence and an awkward statistical position: it examines a sample, not the batch, and its power to detect low-level contamination is limited. That is why how the test is performed, and what happens when it grows something, matters more than the headline price. This page covers choosing a laboratory and reading the result honestly.

1. **Do the method suitability test first**: Before any batch is tested, the laboratory has to show that your product does not inhibit growth of the specified organisms under the test conditions. That validation is product-specific, it takes time, and a laboratory quoting per-batch testing without it has quoted for half the work.
2. **Prefer membrane filtration where the product allows**: Filtration removes the product from the organisms before incubation, which handles antimicrobial products far better than direct inoculation. Direct inoculation is for products that cannot be filtered, and it constrains how much inhibition can be washed away.
3. **Ask where the test is physically performed**: An isolator gives a much lower false positive rate than a cleanroom with an operator in a gown, because most sterility test failures are contamination introduced during the test. Ask for the laboratory's invalid test rate and how it is investigated.
4. **Plan around the incubation period**: The compendial incubation is fourteen days, and that sits on the critical path to release. Build it into the schedule rather than discovering it, and agree how and when results are communicated.
5. **Agree the investigation path before a positive occurs**: A growth positive triggers an investigation into whether it was a genuine batch failure or a laboratory contamination, usually including organism identification and a review of environmental monitoring. Agreeing the process in advance is far better than negotiating it under pressure.

**What is method suitability and why does it come first?**

A product-specific validation showing the product does not inhibit growth of the test organisms under the conditions used. Without it, a negative result could simply mean the product suppressed whatever was there.

**Membrane filtration or direct inoculation?**

Filtration wherever the product can be filtered, because it separates the product from any organisms before incubation and handles antimicrobial products far better. Direct inoculation is the fallback for products that cannot be filtered.

**How long does a sterility test take?**

Fourteen days of incubation under the compendial method, plus method suitability up front for a new product and any investigation if something grows. Rapid methods exist and require their own validation against the compendial method.

**Does a passing sterility test prove a batch is sterile?**

No, and that is not a technicality. It examines a sample and has limited power to detect low-level contamination, which is why sterility assurance rests on the validated process, the environmental monitoring and the media fills, with this test as a confirmation.

**What does sterility testing of pharmaceuticals actually cover?**

The compendial test on a defined number of units, incubated in two media for a stated period, preceded by a method suitability exercise proving the product does not inhibit growth. It covers those units and that period, which is why release depends on the process and the environmental data as much as on the test, and why a passing result is evidence rather than proof.

## Reagent bottles

A reagent bottle looks like the least consequential purchase in the laboratory and then leaches, cracks in an autoclave or loses its contents through a liner that dissolved. Material, closure and light protection are the three decisions, and labeling is the legal obligation that arrives the moment anything is poured into it. This page covers all four.

1. **Choose the material from what goes in it**: Borosilicate glass resists solvents and thermal shock and is the default for organic chemistry and for anything autoclaved. High density polyethylene and polypropylene are unbreakable and suit aqueous solutions, with polypropylene tolerating autoclaving. Fluoropolymer bottles handle aggressive chemistry at a much higher cost.
2. **Pick the closure and check its liner**: The cap is where compatibility usually fails. A polypropylene cap with an inappropriate liner will swell, dissolve or leach into the contents, and for volatile or aggressive material a fluoropolymer-lined closure is what keeps the bottle sealed. Read the liner specification, not the cap material.
3. **Use amber or a sleeve where light matters**: Photosensitive reagents degrade in a clear bottle on an open shelf faster than most people expect. Amber glass or an opaque overpack is cheap, and moving a stock solution into the dark is the most effective stabilisation available.
4. **Match the neck and the thread to the accessories**: Thread sizes are standardized, and a bottle that will take a dispenser, a cap with tubing ports or a safety cap for solvent has to match it. Buying a bottle that cannot take the dispenser you own is an avoidable annoyance repeated at scale.
5. **Label it the moment anything goes in**: Any container holding a hazardous chemical outside immediate use carries the identity, the hazards and the pictograms under the hazard communication rule, plus the date and the person who prepared it for laboratory practice. An unlabelled bottle is the single most common laboratory inspection finding.

**Glass or plastic reagent bottles?**

Borosilicate glass for solvents, for anything autoclaved and where leaching must be minimal. Polyethylene or polypropylene for aqueous solutions where breakage is the bigger risk, with polypropylene where autoclaving is needed.

**Why did my bottle cap fail?**

Almost always the liner rather than the cap. Liners are made from different polymers with very different compatibility, and a liner attacked by the contents leaks, contaminates and can swell enough to jam the closure.

**Do I need amber bottles?**

For photosensitive reagents, yes, and for anything stored on an open bench for months it is cheap insurance. The alternative is a cupboard, which works just as well if the bottle actually stays in it.

**What has to be on the label?**

The chemical identity and the hazard information with pictograms, under the hazard communication rule, for any container other than one in immediate use by the person who filled it. Adding the concentration, the date and the preparer's initials is ordinary good practice.

## Lab safety equipment

Safety equipment is bought as a list and required as a system: the regulation obliges a written plan, the plan decides what is needed, and the equipment is only compliant if it is tested and if people can reach it. This page covers what the laboratory standard actually asks for and the maintenance that most often turns a fitted item into a non-functioning one.

1. **Start from the written chemical hygiene plan**: The laboratory standard requires a written plan naming the standard operating procedures, the control measures, the equipment that must function properly and the circumstances requiring prior approval. Equipment chosen without that plan is a shopping list rather than a control.
2. **Do the hazard assessment before choosing protection**: Eye protection, gloves and clothing are selected from an assessment of the actual hazards, not from habit. Glove material in particular is chemical-specific: a nitrile glove suited to one solvent is permeated quickly by another, and breakthrough time is the number that matters.
3. **Site drench equipment where it can be reached with closed eyes**: Emergency eyewash and drench facilities must be immediately available in the work area, conventionally within about ten seconds' unobstructed travel. Somebody with a chemical splash to the eyes cannot navigate around an obstruction, so the path matters as much as the distance.
4. **Test the drench equipment on a schedule and record it**: Plumbed eyewashes are activated weekly to clear the line and confirm flow, and inspected annually against the performance standard. An untested eyewash delivers stagnant water, which is a second injury on top of the first.
5. **Certify the engineering controls annually**: Fume hoods, biosafety cabinets and the ventilation they depend on are certified on a schedule, with the results posted on the unit. A hood with an expired certificate is not a control, whatever its face velocity reads on the day.

**What chemistry lab safety equipment does the standard actually require?**

It requires a written chemical hygiene plan, that control measures and protective equipment appropriate to the assessed hazards are provided and function properly, and that people are trained. The specific items follow from your own plan and assessment.

**How close does an eyewash have to be?**

Immediately available in the work area, which is conventionally read as about ten seconds of unobstructed travel. The obstruction matters as much as the distance, because somebody with a chemical splash cannot see the route.

**How often should an eyewash be tested?**

Activated weekly to flush the line and confirm flow, and inspected annually against the performance standard. Record both, because the record is what demonstrates the control was real.

**How do I choose gloves?**

By the chemicals in use, from permeation and breakthrough data, not by the box on the shelf. A glove is rated against specific chemicals for a specific time, and the commonest mistake is treating one glove type as general protection.

**What comes before buying protective equipment?**

Asking whether the hazard can be eliminated, substituted or enclosed. Engineering controls beat administrative ones, and consumable protection is the last line rather than the first item on the order.

**Does a clean-air bench protect the operator?**

No. It blows filtered air across the work toward the operator, protecting the sample and not the person. For anything hazardous the answer is a biological safety cabinet or a chemical enclosure.

## NGS pricing

Sequencing quotations are hard to compare because providers price different things: some quote per sample all in, some per lane, some per gigabase with library preparation separate. The headline figure is rarely the number that decides the project cost. This page sets out what to ask so that two quotations answer the same question.

1. **Separate library preparation from sequencing**: These are different operations with different costs and they scale differently. For a small project the library preparation frequently dominates; for a large one the sequencing does. A quotation that bundles them without stating the split cannot be compared with one that separates them.
2. **Fix the unit: per sample, per lane or per gigabase**: A per-lane price is only comparable to a per-sample one once you know how many samples share the lane and at what depth. Convert every quotation to your own unit, which is cost per sample at your required depth, before comparing anything.
3. **State the depth and the read configuration in the request**: Depth, read length and paired or single end change the price substantially. Asking three providers for a price without specifying these produces three quotations for three different experiments.
4. **Ask what happens when a library fails**: Quality control failure rates are real and the policy differs: some providers remake at their cost, some charge, some sequence anyway and report. This clause changes the expected cost of a project more than most line items.
5. **Settle data delivery, retention and analysis**: Which files, in what format, delivered how, retained for how long, and what any bioinformatics costs. Raw data storage and a basic analysis are often assumed to be included and often are not.

**Why do two ngs sequencing price quotations differ so much?**

Usually because they cover different things: one may include library preparation, quality control and analysis, the other may be sequencing alone. Convert both to cost per sample at your required depth with everything you need included, and the gap normally shrinks.

**Is sequencing or library preparation for next generation sequencing the bigger cost?**

For small projects, usually the library preparation, because it is per sample and labor intensive. For large projects at depth, the sequencing dominates. Which side you are on changes which provider is cheaper.

**Should I ask for a price per gigabase?**

It is a useful normaliser between platforms, and it is not what you pay. Translate it into cost per sample at the depth your experiment needs, including library preparation, and compare on that.

**What is usually left out of an NGS cost quotation?**

Failed library remakes, data storage beyond a short window, anything past primary analysis, and shipping. Each is small individually and together they routinely move a project's real cost by a noticeable margin.

## Chromatography supplies

Chromatography stops for consumables far more often than it stops for instruments, and the consumables that stop it are cheap and have long lead times at the worst moment. This page covers what is worth standardising across a laboratory, what belongs on the shelf, and where the quality of a consumable genuinely changes a result.

1. **Standardize fittings, and understand why they are not universal**: Ferrule geometry and the depth the tubing seats differ between manufacturers, and a fitting seated at the wrong depth leaves a void at the column inlet that shows as a split or tailing peak. Pick one system and keep the odd adaptors labeled and separate.
2. **Standardize vials, caps and septa across the autosampler fleet**: Vial dimensions, cap type and septum material are where an autosampler jams or a sample evaporates. One format across the laboratory means one stock line, and septa in particular should be chosen for the solvent rather than for price.
3. **Keep the consumable spares that cause downtime**: Pump seals, check valves, an injector rotor seal, inline filters and a guard column of each method's type. None is expensive, all of them stop work when missing, and all of them have lead times.
4. **Buy solvent at the grade the detector needs**: Chromatography grade solvent matters most with ultraviolet detection at low wavelengths and with mass spectrometry, where trace impurities produce baseline noise and ion suppression. Using a lower grade to save money on a sensitive method is a false economy measured in repeated runs.
5. **Filter everything, sample and mobile phase alike**: Blocked frits are the commonest cause of a dead column, and they arrive from unfiltered samples and from particulates in mobile phase. Syringe filters and mobile phase filtration cost very little against a column.

**Are chromatography fittings interchangeable between brands?**

No. Ferrule geometry and seating depth differ, and a mismatched fitting leaves a dead volume at the column inlet. The peak shape it produces is usually blamed on the column rather than on the fitting.

**Which chromatography supplies should always be on the shelf?**

Pump seals, check valves, an injector rotor seal, inline filters, a spare guard column per method and enough vials and septa for a week. That small set covers most of what stops a laboratory unexpectedly.

**Does solvent grade really matter?**

On sensitive methods, a great deal. Trace impurities raise the baseline under low wavelength ultraviolet detection and suppress ions in mass spectrometry. On a robust method at higher wavelengths the difference is often invisible.

**How long does a guard column last?**

Until the back pressure rises or the peak shape degrades, which depends entirely on how dirty your samples are. Change it on the evidence rather than on a calendar, and keep a spare so the change is not a decision.

**Can I store solvents in an ordinary refrigerator?**

No. A domestic refrigerator has ignition sources inside the compartment. Flammable storage needs a cabinet rated for it, and a system run cold with solvents in it needs the same consideration.

## RNA extraction kits

An RNA preparation is judged on integrity, purity and the absence of genomic DNA, and yield is the least important of the four things people measure. A kit that gives more RNA of lower integrity is the worse kit. This page covers matching the chemistry to the sample, and the two steps that most often decide whether the downstream experiment works.

1. **Choose the chemistry from the sample type**: Silica column kits are fast and clean for cultured cells and soft tissue. Magnetic bead kits automate and suit high throughput. Phenol-based phase separation still wins on difficult, fatty or fibrous tissue and on very small inputs, at the cost of handling a hazardous reagent in a hood.
2. **Homogenise completely, in a lysis buffer that inactivates nucleases**: Incomplete disruption is the commonest cause of low yield and degraded RNA from tissue. Use a mechanical homogeniser appropriate to the tissue, keep everything cold, and get the sample into a denaturing lysis buffer as fast as possible.
3. **Remove genomic DNA and prove that you did**: An on-column or in-solution DNase step removes contaminating DNA, and a no-reverse-transcription control on the downstream assay is what proves it worked. Assuming the step worked is how a quantitative PCR result becomes a DNA measurement.
4. **Measure integrity, not just concentration**: A capillary electrophoresis integrity number on the extract itself, not on an aliquot taken earlier. Purity ratios and concentration say nothing about whether the RNA is intact, and library preparation and reverse transcription both care a great deal.
5. **Handle the awkward sample types deliberately**: Archival fixed tissue gives short, chemically modified RNA and needs a dedicated kit and a library method that tolerates it. Blood needs globin considerations. Plant and bacterial samples need their own lysis. A general kit used on these will disappoint quietly.

**Which rna extraction kit should I use?**

Column kits for cultured cells and soft tissue, bead kits for automation and throughput, phenol-based methods for difficult or fatty tissue and very small inputs. The sample decides, not the throughput you would like.

**Do I need a DNase step?**

Whenever the downstream assay cannot distinguish RNA from DNA, which includes most quantitative PCR where the primers do not span an intron. Run a no-reverse-transcription control to prove the step worked rather than assuming it.

**What integrity number is good enough?**

It depends on the application: library preparation for standard RNA sequencing wants high integrity, while a short amplicon assay tolerates far more degradation. Measure it, record it and set the threshold from the method you will actually run.

**Why is my yield fine and the downstream experiment failing?**

Usually a co-purifying inhibitor or degradation that the absorbance reading cannot see. Check the integrity, check the two hundred and thirty ratio for organic carryover, and try a dilution series in the downstream assay, which exposes inhibition quickly.

**Does a total RNA extraction kit keep small RNAs?**

Only if it is a total protocol. Columns optimized for messenger RNA lose most of the short fraction at binding, and it cannot be recovered afterwards, so the choice has to be made before extraction.

**Do I need a DNase kit to digest the DNA?**

If any amplification target could be amplified from genomic DNA, yes, and a control without reverse transcriptase is what proves it worked. Remove or inactivate the enzyme afterwards.

## Biology lab equipment

A new laboratory usually gets its equipment budget in one block and spends it in the wrong order, buying the instrument that does the exciting experiment before the infrastructure that makes any experiment possible. This page sets out what has to be in the room before work starts, what can be shared, and what is better bought as a service until the science is proven.

1. **Containment and culture first**: A certified biosafety cabinet and a carbon dioxide incubator decide whether cell work can happen at all, and both have lead times and installation requirements. Nothing else in a cell biology laboratory matters until these are in and certified.
2. **Then the separations you use every day**: A benchtop centrifuge and a microcentrifuge, refrigerated if you work with protein or RNA. These are used constantly by everybody, and a laboratory with one shared microcentrifuge develops a queue within a week.
3. **Then amplification and detection**: A thermal cycler, a quantitative instrument if your work needs one, and a way to read gels or blots. Buy the throughput you actually have rather than the throughput you hope for, since these are the easiest items to add later.
4. **Then cold storage, sized with headroom**: A laboratory refrigerator, a minus twenty freezer and a minus eighty with room to grow, plus monitoring and alarms from day one. Cold storage is where samples are lost and it is the item most often bought exactly to current need.
5. **Buy imaging, sequencing and cytometry as a service until the work is proven**: A microscope, a sequencer and a cytometer are the instruments most often bought too early and used least. Core facility access gives better instruments with expert support, and the purchase argument becomes obvious once scheduling is the constraint.

**What common laboratory equipment does a new biology laboratory need first?**

A certified biosafety cabinet, a carbon dioxide incubator, refrigerated benchtop and microcentrifuges, a thermal cycler, cold storage with alarms, a balance, a pH meter, a water purification system and a working pipette set.

**What biology lab equipment is better shared?**

Anything used intermittently and expensive to own: confocal and high content imaging, sequencing, cytometry, mass spectrometry and ultracentrifugation. Core access usually gives better instruments and expertise for less than ownership.

**How much cold storage should a new laboratory buy?**

More than current need, with the failure plan written before the first sample goes in. Freezers fill faster than anyone predicts and the day a full freezer fails is the day you find out whether there was anywhere to put its contents.

**New or refurbished?**

Refurbished works well for mechanically simple items such as centrifuges, incubators and shakers, from a supplier offering a warranty and service. For anything with optics, detectors or a regulated role, the service history matters more than the saving.

## LIMS for a small lab

A small laboratory buying a system designed for a large one ends up with an implementation project instead of a system. The features that justify an enterprise product are the same features that make it slow to configure, and a laboratory of six people rarely has anyone whose job that can be. This page covers what to look for and what to decline.

1. **Prefer a hosted, web based system unless there is a reason not to**: A web based lims removes the server, the operating system patching, the database administration and the backup regime, none of which a small laboratory has staff for. The trade is dependence on the provider's availability and a subscription rather than a capital purchase.
2. **Count the configuration effort, not the feature list**: Ask how long implementation takes for a laboratory your size, who does it, and what it costs. A product configured by a consultant over three months is a different purchase from one a laboratory manager sets up in a fortnight.
3. **Decline the regulatory tier you do not need**: Electronic signature workflows, full audit trail review and validation packages are necessary for regulated work and pure cost for a research laboratory. Buy the tier your obligations require and no more.
4. **Check the instrument interfaces you will actually use**: One or two integrations usually cover most of a small laboratory's volume. Ask whether those specific instruments are supported out of the box, because a bespoke interface is a disproportionate cost at this scale.
5. **Make adoption the deciding test**: Have two of your own staff enter a day's real work during the trial. A system nobody uses is worse than a spreadsheet, and at this size one reluctant user is a large fraction of the laboratory.

**What should I look for in LIMS for small labs?**

Fast implementation, a hosted option, sensible pricing per user, the instrument interfaces you actually use, and an interface your staff will adopt. Feature breadth matters far less than whether it is configured and used.

**Is a web based lims a good idea for a small laboratory?**

Usually yes, because it removes the server, the backups and the patching, which a small laboratory has nobody to do. Check the availability commitment, where the data is held and how you would get it out.

**Could a spreadsheet do the job?**

For a very small operation, sometimes, until two people edit it at once or somebody asks what a result was six months ago. The moment traceability or concurrency matters, a spreadsheet stops being cheaper.

**How long should implementation take?**

Weeks rather than months at this size. If a vendor's answer is a multi-month project with a consultant, they are quoting their enterprise product and you are being sold the wrong thing.

## LIS, LIMS and SDMS

These acronyms are used interchangeably in procurement documents and they describe genuinely different systems built around different central objects. A clinical laboratory information system is organized around a patient and an order; a laboratory information management system is organized around a sample and a test; a scientific data management system is organized around the raw instrument file. Buying the wrong one is an expensive mistake that takes a year to become obvious.

1. **A LIS is built around the patient and the order**: It receives orders from a clinical system, holds patient demographics, produces a patient-oriented report and interfaces to the electronic health record, conventionally over HL7. Its world is people and episodes of care, and its regulatory context is clinical.
2. **A LIMS is built around the sample and the test**: It accessions a sample, routes it through methods and batches, holds the quality control around those batches, and reports against a specification. Its world is specimens and analyses, and it serves research, testing and manufacturing laboratories rather than clinical care.
3. **An SDMS is built around the raw file**: It captures the instrument's raw output, indexes it, keeps it in a controlled repository and makes it retrievable years later regardless of which software version produced it. It is the answer to the question of where the actual chromatogram lives when the chromatography data system is retired.
4. **An ELN is built around the experiment**: It records what a person did and why, in narrative form, with attachments. It overlaps with a LIMS at the edges and is the wrong place to hold a structured result set, which is the commonest reason a laboratory ends up with both.
5. **Decide from the central object, not the feature list**: Ask what the thing your laboratory tracks actually is. If the answer is a patient, you want a LIS. If it is a sample, a LIMS. If it is a file, an SDMS. A vendor can usually configure any of them to do a little of the others, and that configuration is where projects run over.

**What is the difference between LIS and LIMS?**

A laboratory information system is patient-centred and built for clinical diagnostics, interfacing with the electronic health record. A laboratory information management system is sample-centred and built for testing, research and manufacturing laboratories reporting against specifications.

**Can one system do both?**

Some vendors offer both, and in practice the clinical and the non-clinical workflows are usually served by separate configurations even within one product. A laboratory doing both kinds of work should expect two configurations rather than one compromise.

**SDMS, LIMS or both: do I need an SDMS as well?**

If raw instrument data has to be retrievable for years independently of the acquisition software, yes. Many laboratories discover this at the point a chromatography system is replaced and the old raw files become unopenable.

**Where does an ELN fit?**

Alongside, recording what was done and why. Keeping structured results in an ELN and narrative in a LIMS is the arrangement that causes the most trouble, and it is surprisingly common.

## Cloud LIMS

Hosting is an operational decision dressed up as a technology one. Moving a laboratory system to a vendor's infrastructure removes a set of tasks most laboratories have nobody to do, and it introduces a dependency and a contract that last as long as the system does. This page sets out what actually changes and the four questions worth settling before signing.

1. **The vendor takes the infrastructure work**: Server provisioning, operating system patching, database administration, backup and restore and availability monitoring stop being yours. For a laboratory with no dedicated information technology staff that is the whole argument, and it is usually a sufficient one.
2. **You keep the configuration and the validation**: Hosting does not configure your workflow or validate your system for its intended use. Where the work is regulated, the validation obligation remains with you, and the vendor's own qualification documentation is an input to it rather than a replacement for it.
3. **Availability becomes a contract term, not a control**: When the system is down you wait. Read the availability commitment, what it excludes, what maintenance windows look like and what the remedy is, and decide whether the laboratory can work on paper for that long.
4. **Data location and access become questions you must answer**: Where the data physically sits, who at the vendor can see it, how it is encrypted and what happens in a legal request are questions a regulated or clinical laboratory will be asked. Get the answers in writing before, not during, an audit.
5. **Exit is the term nobody negotiates and everybody needs**: What an export contains, in what format, whether it includes the audit trail, how long data is retained after termination and what assistance is included. Test the export during the trial rather than trusting the answer.

**Is a cloud based lims suitable for regulated work?**

Yes, routinely, provided the vendor can supply its qualification documentation and you perform validation for your intended use. Hosting does not transfer the regulatory obligation, and a vendor claiming it does is worth questioning.

**What does a cloud based lims system remove from my workload?**

Server provisioning, patching, database administration, backups and availability monitoring. It does not remove configuration, user administration, validation or training, which are the larger part of the effort in most implementations.

**What happens if the internet connection fails?**

The laboratory stops using the system. Ask what the documented fallback is, keep a paper or local procedure for the critical path, and test it, because a connection failure at a busy moment is when the procedure is needed and never rehearsed.

**Can I get my data out of a LIMS cloud service?**

Only as well as the contract and the product allow. Ask for a full export including the audit trail in an open format, and run it during the evaluation. This is the single most useful hour of a hosted software trial.

## Biobanking LIMS

A biobank is not a testing laboratory with a big freezer. Its central object is a stored aliquot with a lineage, a consent and a physical position, and it usually performs very few tests. A general laboratory system configured for a biobank spends its life fighting a data model built around analyses that mostly are not happening. This page covers what the system actually has to do.

1. **Aliquot lineage, from collection to the vial in your hand**: A parent specimen becomes aliquots, which become derivatives, which are sometimes re-aliquoted. The system must hold that tree, so that a question about one vial can be answered about its parent, and so that a consent withdrawal reaches every descendant.
2. **Position to the box and the slot**: Freezer, rack, box and position, with the box map visible. Searching a minus eighty freezer with the door open destroys the samples around the one you are looking for, and a system that stops at the freezer number has not solved the problem it was bought for.
3. **Consent scope against every sample, not against the study**: Which uses are permitted, under which approval, until when. The question asked of a biobank is always about specific samples and a specific proposed use, and a system that cannot answer it forces a conservative refusal.
4. **Temperature history beside the inventory**: An excursion affects the samples that were in that unit at that time, so the monitoring record and the inventory have to be joinable. Held in separate systems, an excursion becomes a manual reconstruction at exactly the wrong moment.
5. **Request, approval and distribution as a workflow**: Biobanks issue material against approved requests, and the record of who asked, who approved, what was shipped and what is left is the audit. Handling that in email and a spreadsheet is where biobank inventories stop matching reality.

**Why not use a general sample management LIMS for biological sample management?**

Because a general system is built around tests and a biobank is built around stored aliquots, their lineage, their consent and their position. Everything works, awkwardly, and the awkwardness compounds at scale.

**How detailed does LIMS sample tracking need to be?**

To the position in the box. Anything coarser means searching with the freezer door open, which harms the samples around the target, and it makes a physical audit impossible to complete.

**What happens when a donor withdraws consent?**

Every sample descended from that donation has to be findable and actioned, which is exactly why lineage has to be held as data. A system that cannot walk the tree makes withdrawal an unverifiable manual exercise.

**Should the biobank system hold the analytical results too?**

Usually it holds what was measured on a sample and points at the analysis, rather than being the analytical system. Keeping the biobank's job to custody and the laboratory's job to testing keeps both data models honest.

## Phase contrast

Phase contrast is what makes unstained living cells visible, and it is bought as an option that is then never aligned. The technique needs a ring in the condenser and a matching ring in the objective, correctly superimposed, and an instrument where those two are out of alignment gives a dim, haloed image that people put up with for years. This page covers specifying it and setting it up.

1. **Understand what it converts**: Living cells barely absorb light, so they are nearly invisible in brightfield. They do shift the phase of light passing through them, and phase contrast converts that shift into an intensity difference. That is why it is the default for anything in a culture vessel.
2. **Buy the condenser annulus and the objective as a matched pair**: Each phase objective needs its own annulus size in the condenser, and the sizes are not universal between manufacturers. A phase objective in a stand whose condenser has no matching ring produces a dark, low contrast image and nothing can be adjusted to fix it.
3. **Align the rings with a centring telescope, then check it regularly**: The annulus image has to sit exactly over the objective's phase ring, viewed through a centring telescope or a Bertrand lens. This takes a couple of minutes and is the difference between a good phase image and the dim haloed one most instruments produce.
4. **Expect a halo and know what it is**: A bright halo around edges is inherent to the method, not a fault. It obscures fine detail at boundaries, which is where differential interference contrast does better, at higher cost and with polarisation optics that plastic vessels disturb.
5. **Check vessel compatibility before buying**: Phase contrast works through plastic culture vessels, which is its practical advantage over differential interference contrast. Thick vessel bottoms and condensation still degrade it, so match the working distance and the condenser to the vessels you actually use.

**What does a phase contrast microscope show that brightfield does not?**

Unstained, transparent specimens. Living cells in culture have almost no absorption contrast and are effectively invisible in brightfield; phase contrast converts the phase shift they cause into visible intensity differences.

**Why is my phase contrast image dim and washed out?**

Almost always the annulus is not aligned with the objective's phase ring, or the wrong annulus is in the condenser for that objective. Align with a centring telescope and check the ring size matches the objective.

**Phase contrast or differential interference contrast?**

Phase contrast for plastic culture vessels and for routine live cell observation. Differential interference contrast for a shallower optical section and better edge detail on glass, at greater cost and with polarisation optics plastic disturbs.

**Can I add phase contrast to an existing microscope?**

If the condenser accepts annuli and the turret takes the objectives, usually yes. Check whether the condenser is a phase-capable model first, because adding phase objectives to a plain condenser achieves nothing.

## Atomic force microscopy

An atomic force microscope does not image with light or electrons; it feels the surface with a sharp tip on a flexible cantilever and builds a map of height. That makes it the instrument that gives true three dimensional topography, works in liquid on living material, and measures mechanical properties, and it makes the probe and the mode the decisions that matter. This page covers those.

1. **Choose the mode from the sample's fragility**: Contact mode drags the tip across the surface and damages soft material. Tapping or intermittent contact oscillates the cantilever and touches briefly, which is what makes biological samples imageable. Force spectroscopy holds position and measures the force curve, which is how stiffness is obtained.
2. **Match the cantilever spring constant to the sample**: A stiff cantilever on a soft cell deforms it; a soft cantilever on a hard surface gives no contrast. The spring constant is the single most consequential consumable choice, and it has to be calibrated on the instrument rather than taken from the box.
3. **Decide whether you need liquid operation**: Imaging in buffer is what allows living cells and hydrated biomolecules to be measured in a meaningful state. It requires a liquid cell, a different cantilever holder and careful attention to drift, and it is not a trivial add-on.
4. **Prepare the sample so it stays put**: Whatever you image must be firmly attached to a flat substrate, because the tip will otherwise push it around. Mica and functionalised glass are the usual choices, and immobilisation chemistry is most of the method development on a new sample type.
5. **Site it away from vibration and acoustic noise**: Vertical resolution is below a nanometre, so building vibration, footsteps and air currents all appear in the image. An acoustic enclosure and an isolation table are part of the instrument rather than accessories, and a basement room beats a fifth floor one.

**What does an afm microscope measure?**

Surface height, by touching the sample with a sharp tip on a flexible cantilever and tracking the deflection. It gives true three dimensional topography and, in force spectroscopy mode, mechanical properties such as stiffness and adhesion.

**Can it image living cells?**

Yes, in buffer, in a gentle oscillating mode with a soft cantilever. It is one of the few techniques that measures the mechanics of a living cell, and it is slow compared with optical imaging, which is the trade.

**How does it compare with electron microscopy?**

An electron microscope gives higher lateral resolution and needs vacuum and usually a coating. Atomic force microscopy gives true height data, works in air or liquid on unstained material, and has far lower lateral resolution. They answer different questions.

**What limits the resolution in practice?**

The tip radius laterally, and vibration vertically. A blunt or contaminated tip is the commonest cause of a disappointing image, and tips are consumables that degrade during use rather than lasting.

## HPLC consumables

Chromatography stops for consumables far more often than for instruments, and the items that stop it are inexpensive and have lead times at the worst possible moment. This page covers what is worth holding, where consumable quality genuinely changes a result, and the two items that cause most autosampler and column failures.

1. **A guard column of each method's chemistry**: A guard column takes the particulates and the strongly retained material that would otherwise sit on the analytical column's inlet frit. It costs a fraction of the column and is changed on evidence, which means a rising back pressure or a degrading peak shape rather than a calendar.
2. **Vials and septa matched to the autosampler and the solvent**: Vial dimensions, cap type and septum material are where an autosampler jams, a needle coring occurs or a sample evaporates. Choose the septum for the solvent rather than for price, and standardize one format across the laboratory so there is one stock line.
3. **Filters, on both the sample and the mobile phase**: Blocked inlet frits are the commonest cause of a dead column and they arrive from unfiltered samples and particulates in mobile phase. Syringe filters of a compatible membrane, and mobile phase filtration, cost very little against a column.
4. **Pump seals, check valves and a rotor seal**: These are consumables rather than spare parts, and their failure presents as pressure ripple, drifting retention times and carryover. Holding one set of each turns a day of downtime into twenty minutes.
5. **Solvent at the grade the detector needs**: Trace impurities matter most with low wavelength ultraviolet detection and with mass spectrometry, where they raise the baseline and suppress ions. Using a lower grade on a sensitive method is a false economy paid for in repeated sequences.

**Does a guard column hplc setup actually extend column life?**

Substantially, on dirty samples. It collects the particulates and strongly retained material that would otherwise foul the analytical column's inlet, and it costs a small fraction of the column it protects.

**Do hplc vials really matter?**

They cause a disproportionate share of autosampler failures. Dimensions outside tolerance jam the tray, the wrong septum cores and drops fragments into the sample, and a poor seal lets volatile solvent evaporate before injection.

**Which hplc consumables should always be on the shelf?**

A guard column per method, vials and septa for a week, syringe filters, pump seals, check valves and an injector rotor seal. All are inexpensive and all of them stop work when missing.

**How often should a guard column be changed?**

On evidence rather than on a schedule: when back pressure rises or peak shape degrades. Keep a spare so the change is a two minute decision rather than an order.

**Why did a new peak appear after I changed vials?**

Most often the septum. Closures release extractables under some solvents, and glass type changes adsorption of basic compounds. Fix vial, cap and septum as part of the method and treat a consumable change as a method change.

## GC columns

A gas chromatography column is specified by four things and most selection errors are in the first of them. Phase polarity decides what separates; film thickness decides capacity and retention of volatiles; internal diameter decides efficiency and sample capacity; length decides resolution and run time. This page covers each and the installation step that ruins more new columns than anything else.

1. **Match the stationary phase polarity to the analytes**: Non-polar phases separate largely by boiling point and suit hydrocarbons and most general screening. Polar phases add selectivity for alcohols, acids and other polar compounds. Choosing the phase is choosing the separation, and everything else only refines it.
2. **Set film thickness from the volatility of what you are separating**: Thicker films retain volatile compounds long enough to separate them and increase sample capacity, at the cost of longer run times and more bleed at high temperature. Thin films suit high boiling compounds and give faster runs.
3. **Choose the internal diameter from capacity against efficiency**: Narrower columns are more efficient and carry less sample before overloading. A quarter millimetre bore is the general purpose default; go narrower for resolution with a clean concentrated sample, wider where sample capacity or robustness matters.
4. **Pick length last, and be sceptical of long columns**: Doubling length gives roughly a forty percent gain in resolution and doubles the run time. Changing phase or optimizing the temperature program usually buys more separation than length does, and far more cheaply.
5. **Install and condition it properly, every time**: Cut the column end square with a proper cutter, set the insertion depth into the inlet and detector to the manufacturer's specification, and condition under carrier flow before the first run. A ragged cut or a wrong insertion depth produces peak tailing that is then blamed on the column.

**How do I choose a gas chromatography column phase?**

From the polarity of your analytes. A non-polar phase separates broadly by boiling point and covers most general work; a polar phase adds selectivity for alcohols, acids and similar compounds. The phase choice is the separation.

**Does a longer column always give better separation?**

No. Resolution rises with the square root of length, so doubling the column gives about forty percent more resolution and doubles the run time. Changing the phase or the temperature program is usually the better move.

**What is column bleed and when does it matter?**

Stationary phase degrading at high temperature and reaching the detector, which raises the baseline and, with mass spectrometric detection, adds characteristic ions. Low bleed columns are worth the premium for trace work and for mass spectrometry.

**Why does a new column give tailing peaks?**

Usually installation: a ragged cut, the wrong insertion depth in the inlet or detector, or a leaking ferrule. Check those three before concluding the column is faulty, because they account for most new-column complaints.

**Can I use an ordinary column with a mass detector?**

Often, but a low bleed column is worth the premium: material leaving the stationary phase enters the source and raises background everywhere. Condition the column before connecting it to the source.

## Silica gel columns

Column chromatography by hand is still the cheapest purification in a synthetic laboratory and it is unforgiving of three things: a badly packed bed, too much material, and a load that is already halfway down the column before the gradient starts. This page covers packing, sizing and loading, and when to reach for an automated system instead.

1. **Choose the mesh from how you will run it**: Coarser silica flows under gravity and gives poorer resolution; the finer flash grade around two hundred and thirty to four hundred mesh needs pressure and gives much better separation. Buying the wrong grade for your setup produces either a column that will not flow or one that separates badly.
2. **Size the bed from the difficulty, not just the mass**: Twenty times the sample mass in silica is generous for a well separated pair; fifty times or more is needed for close-running compounds. A short wide column is fast and resolves poorly; a tall narrow one is the opposite.
3. **Pack a bed with no channels and no air**: Slurry the silica in the starting solvent, pour it in one go, and tap the column while it settles. A dry-packed bed wetted afterwards traps air, and a channel is invisible until the band running down it arrives early and broadened.
4. **Keep the bed covered, always**: A bed that runs dry cracks, and a cracked bed cannot be recovered. Add a sand layer on top so pouring solvent does not disturb the surface, and never leave a running column unattended near the end of a reservoir.
5. **Dry load anything that will not dissolve in a weak solvent**: Liquid loading in a strong solvent pushes the sample down the column before separation begins. Adsorbing the crude onto a small amount of silica and loading it dry gives a tight starting band and fixes most disappointing separations by itself.

**What mesh silica should I use for column chromatography?**

Around 230-400 mesh flash grade where you can apply pressure, which is most modern practice. Coarser grades flow under gravity and separate noticeably worse, and are worth using only when no pressure source is available.

**How much silica do I need?**

Sized to the difficulty of the separation: roughly twenty times the sample mass for an easy one and fifty times or more for close-running compounds. Sizing on mass alone while ignoring the plate is the commonest reason fractions overlap.

**Why did my column separate badly?**

A channel in the bed, an overloaded column, or a load applied in too strong a solvent. Repacking properly, using more silica and dry loading address almost all of it, in that order of likelihood.

**When should I use an automated flash system instead?**

When the separation is repeated, when the record matters, or when the volume of work makes hand collection tedious. An automated system gives a chromatogram, which is evidence about purity that a rack of tubes is not.

## FPLC

Fast protein liquid chromatography is a separate instrument class from analytical HPLC for two reasons that are easy to overlook: the flow path is biocompatible, because stainless steel and proteins interact badly, and the pressure class is low, because the soft resins that separate proteins collapse under pressure. This page covers specifying one and building methods it can run alone.

1. **Insist on a biocompatible flow path**: Titanium and polymer wetted parts rather than stainless steel, because iron surfaces adsorb and can oxidise proteins, and because the high salt buffers used here corrode steel. This is the property that distinguishes the instrument class more than any performance figure.
2. **Match the pressure class to the resins you will use**: Agarose and other soft resins compress at pressures an analytical system generates routinely. A protein system runs at a few bar to a few tens of bar, and a column compressed once does not recover.
3. **Monitor absorbance, conductivity and pH together**: Absorbance at two hundred and eighty nanometres follows the protein; conductivity tells you the gradient was actually delivered; pH tells you the buffer system is doing what the method says. A run followed on absorbance alone hides the two commonest failures.
4. **Build the whole train as one method**: Capture, a loop, and a polishing size exclusion step chained in one method removes the overnight gap where an acidic eluate sits and aggregates. That chaining is the single largest recovery gain automation offers on a protein purification.
5. **Set the safety limits before the first unattended run**: A pressure alarm at the column's specification rather than the pump's, an air sensor on the inlet, and pause rather than continue on alarm. Most overnight column losses are an emptied buffer bottle or a blocked frit that the system pushed through.

**FPLC vs HPLC: what is the difference in protein chromatography?**

Flow path materials and pressure class. Protein systems use biocompatible titanium and polymer wetted parts and run at low pressure to suit soft resins; analytical HPLC uses stainless steel at high pressure for small rigid particles.

**Can I run protein separations on an analytical HPLC?**

For some robust applications on rigid media, yes. For soft resins and for proteins sensitive to metal surfaces, the pressure will compress the bed and the flow path may cost you activity, so the answer is usually no.

**Why monitor conductivity as well as absorbance?**

Because conductivity shows whether the salt gradient was actually delivered. A run that failed because a buffer line ran dry looks like a protein problem on the absorbance trace alone, and the conductivity trace makes it obvious.

**Does the system need to be in a cold room?**

For marginally stable proteins, yes, and many purifications that fail overnight at room temperature succeed cold. A refrigerated cabinet is the alternative where a cold room is not available.

**Can one system purify and analyse protein?**

Poorly. A preparative system is built to recover material intact at moderate pressure with fraction collection; an analytical system is built to measure. A laboratory doing both usually needs both.

## High throughput HPLC

Laboratories trying to raise chromatographic throughput almost always attack the wrong number. The separation is rarely the longest part of a cycle: re-equilibration, injector overhead, needle wash and the time between sequences usually are, and none of them appears in the chromatogram. This page covers where the time actually goes and how to recover it without damaging the method.

1. **Measure the whole cycle, not the run time**: Add the gradient, the hold, the re-equilibration, the injector cycle and the needle wash. On many methods the analyte has eluted in a third of the cycle and everything after it is overhead that nobody has looked at.
2. **Shorten re-equilibration deliberately, then prove it**: Re-equilibration is frequently set far longer than the column needs. Shorten it in steps and check retention time reproducibility over a full sequence; the last part of the equilibration is usually doing nothing and is the cheapest time available.
3. **Scale the method onto smaller particles rather than truncating it**: Moving to sub-two-micron packing with the column dimensions, flow and gradient scaled together preserves the separation at a fraction of the run time. Simply steepening a gradient to save minutes loses the resolution the method was validated on.
4. **Overlap the injector with the run where the system allows**: Preparing the next injection during the current run removes the injector cycle from the critical path. On a system that supports it, this is free throughput and it is frequently switched off.
5. **Fix the sample preparation, which is usually the real ceiling**: A two minute analysis fed by a twenty minute extraction is a twenty minute method. Plate-based extraction, automation and parallel preparation move the ceiling far more than any instrument change.

**How do I increase high throughput hplc output?**

Measure the whole cycle first. Re-equilibration, injector overhead and needle wash usually dominate, and recovering them costs nothing. After that, scaling onto smaller particles preserves the separation while shortening the run.

**Can I just shorten the gradient?**

Not without changing the separation. Where the method is registered, arbitrary gradient changes may exceed the allowed adjustment range and become a modification requiring revalidation. Scale the method properly instead.

**Is a second instrument cheaper than optimising?**

Sometimes, and it is the honest comparison to make. A second system also adds a second set of maintenance, qualification and a second queue to manage, so the optimisation is usually worth doing first.

**What limits hplc analysis throughput in most laboratories?**

Sample preparation and data review, not the chromatography. Automating extraction and using processing methods that do not need manual integration on every peak usually buys more than any instrument change.

## Western blot supplies

Transfer is the step that loses proteins, and most of the loss is decided by consumables rather than by settings. Membrane chemistry and pore size, the thickness and cut of the blotting paper, and the compression a cassette still applies are the three things that decide whether the sandwich makes even contact. This page covers all three and what is worth holding in stock.

1. **Choose the membrane chemistry for the detection and the target**: Nitrocellulose binds well, has low background and is fragile. Polyvinylidene fluoride is mechanically robust, binds more protein, requires methanol wetting and is the usual choice for fluorescent detection and for stripping and reprobing.
2. **Pick the pore size from the protein size**: A nought point four five micrometre pore suits most proteins; a nought point two micrometre membrane retains small proteins and peptides that otherwise pass straight through. A small target lost in transfer is very often a pore size choice nobody revisited.
3. **Use thick blotting paper, cut exactly to the gel**: Blotting paper holds the buffer reservoir that carries the current through the sandwich. Thin or unevenly cut paper gives uneven current and patchy transfer, and paper larger than the gel lets current bypass it entirely at the edges.
4. **Retire cassettes and pads when they stop compressing**: Foam pads compress permanently with use and a cassette that no longer applies even pressure gives exactly the patchy transfer people blame on the gel. Date the pads and replace them; they are cheap and they are consumables.
5. **Hold the stock that stops work**: Membrane, thick paper, pads, cassettes and blocking reagent for a fortnight. None is expensive and each of them has stopped a blot on a Friday afternoon somewhere.

**Nitrocellulose or PVDF?**

Nitrocellulose for low background and straightforward chemiluminescence. PVDF for mechanical robustness, higher binding capacity, fluorescent detection and any blot you intend to strip and reprobe. PVDF must be wetted in methanol first.

**Which pore size should I use?**

0.45 micrometre for most proteins, 0.2 micrometre for small proteins and peptides. If a small target is transferring poorly, the pore size is the first thing to change.

**Does western blot paper really matter?**

Yes. It is the buffer reservoir that carries current through the sandwich, so thickness and an exact cut to the gel size decide how evenly the transfer happens. Paper overhanging the gel lets current bypass the sandwich.

**How do I know a western blot transfer cassette is worn?**

Patchy transfer that moves with the cassette rather than with the gel, and pads that no longer spring back. Date the pads when they are first used and replace them on that basis rather than waiting for a bad blot.

**Which membrane should I use?**

Nitrocellulose for low background on a single probe; the fluoropolymer membrane where you need higher binding capacity or intend to strip and reprobe. If a small protein disappears while others transfer, change the pore size first.

**Why do two ladders disagree about a band's size?**

Pre-stained ladders carry dye that shifts their apparent mass and behaves differently between gel systems. Use them to confirm the transfer and bracket the target; where the mass is the claim, run unstained standards.

## 16S sequencing

Amplicon sequencing of the bacterial ribosomal gene is cheap, fast and easy to over-interpret. Its resolution stops around the genus for most regions, its results are compositional rather than absolute, and low biomass samples are dominated by contamination from the kit unless controls were sent. This page covers specifying the work and reading the result honestly.

1. **Choose the variable region for what you need to distinguish**: Different hypervariable regions resolve different taxa, and no single region resolves everything. A commonly used pair of adjacent regions gives broad coverage; a single region gives better sequence quality per read. Ask the provider which primers they use and what that choice cannot see.
2. **Send negative controls, and insist they are sequenced**: Extraction blanks and no-template controls carried through the whole process are how kit and reagent contamination is identified. On low biomass samples, contamination is frequently most of the signal, and without blanks it is indistinguishable from biology.
3. **Send a mock community as a positive control**: A defined mixture of known organisms at known proportions shows what the whole pipeline does to composition. It is the only way to know whether a shift you observe is biology or a bias in extraction, amplification or the classifier.
4. **Fix the extraction, because it biases the answer**: Gram-positive organisms with tough walls need mechanical lysis, and a gentle extraction under-represents them systematically. Use one extraction method for the whole study and describe it, because comparing across extraction methods compares the methods.
5. **Agree the analysis, including the reference database version**: Taxonomic assignment depends on the classifier and the reference database version, and results change when either does. Name them in the methods, and reanalyse a study's samples together rather than in batches across a database update.

**Can a 16s sequencing service identify organisms to species?**

Rarely and unreliably. Short amplicon regions generally resolve to genus, with species assignments trustworthy only in particular groups. Where species matters, full-length sequencing or shotgun metagenomics is the honest route.

**Why do I need blanks on a 16S run?**

Because extraction kits and reagents carry bacterial DNA. On low biomass samples that background can be most of what is detected, and without sequenced blanks there is no way to separate it from the sample.

**Does 16S tell me how much bacteria is present?**

No. The data is compositional: it reports proportions within each sample, not absolute abundance. Pairing it with a quantitative measurement is the only way to make absolute statements.

**16S or shotgun metagenomics?**

16S for cheap, broad community profiling at genus level across many samples. Shotgun where species or strain resolution, functional content or non-bacterial members matter, at considerably higher cost per sample.

## Single cell libraries

Single cell work is expensive per sample and unforgiving of poor input, and the decision that matters most is made before any kit is opened: whether the question needs many cells shallowly or few cells deeply. This page covers that choice, what it shares with low input bulk work, and the sample quality thresholds that decide whether the run produces data or an expensive lesson.

1. **Decide many cells shallowly or few cells deeply**: Droplet chemistries profile thousands to tens of thousands of cells at a transcript end, which suits composition and cell type discovery. Plate based methods profile hundreds of cells with full length coverage, which suits isoform work, low expressed genes and anything needing the whole transcript.
2. **Get the suspension right, because it is the whole experiment**: A single cell suspension with high viability, no clumps and no debris is the requirement. Viability below roughly nine in ten introduces ambient RNA that contaminates every droplet, and dissociation stress itself changes the transcriptome, which is a well documented artefact.
3. **Count and load deliberately**: Loading concentration sets the balance between cells captured and doublets formed. Count accurately on the suspension you will actually load, not on an earlier aliquot, and accept fewer cells rather than a doublet rate that confounds every cluster.
4. **Treat low input bulk as a different problem**: Low input RNA sequencing from a few cells or a biopsy shares the amplification sensitivity but keeps bulk statistics. Where the question is about average expression in a small sample rather than about heterogeneity, low input bulk is cheaper, simpler and better powered.
5. **Plan the analysis before the run**: Ambient RNA correction, doublet detection, the integration method across samples and the clustering resolution all change the answer. Choosing them after seeing the clusters is how single cell studies produce cell types that do not replicate.

**Droplet or plate based sc RNA seq library preparation?**

Droplet for thousands of cells and questions about composition and cell type. Plate based for full length coverage, isoform level questions and small numbers of precious or sorted cells. They answer different questions at very different costs per cell.

**What viability does a single cell sequencing service need?**

As high as you can achieve, and above roughly nine in ten for droplet chemistries. Dead cells release RNA into the suspension, and that ambient RNA is then captured in every droplet and has to be corrected computationally.

**When is low input or FFPE RNA seq the better choice?**

When the question is about average expression in a small sample rather than about heterogeneity between cells. It is cheaper, has better established statistics and does not pay the cost of single cell sparsity.

**How many cells do I need?**

Enough to see the rarest population you care about several times over, which follows from the expected frequency rather than from a round number. And enough biological replicates, which is the constraint single cell studies most often ignore.

## Imaging cytometry

A conventional cytometer reports intensity per event and leaves you inferring what happened inside the cell. An imaging cytometer captures a picture of every event, which settles questions that intensity alone cannot: whether a signal is on the surface or inside, whether two markers are in the same place, and whether the event was one cell at all. This page covers where that is worth its cost.

1. **Internalisation, which intensity cannot distinguish**: A labeled ligand bound to the surface and one taken into the cell give similar total intensity and completely different biology. An image resolves it directly, and this is the application where imaging cytometry most clearly beats the alternatives.
2. **Colocalisation, on thousands of events rather than dozens**: Whether two markers occupy the same compartment is a microscopy question answered on a handful of cells. An imaging cytometer answers it with the statistics of a cytometry experiment, which is a genuinely different kind of evidence.
3. **Nuclear translocation as a measured quantity**: Similarity between a transcription factor's signal and a nuclear stain gives a per-cell translocation score across the whole population, rather than a qualitative impression from selected fields.
4. **Doublets, debris and morphology, settled by looking**: Scatter based doublet exclusion is inference; an image is evidence. For samples where aggregation is a real problem, being able to see what was gated is worth a great deal of argument avoided.
5. **Accept the cost in throughput and sensitivity**: Acquisition is slower than a conventional analyser and detector sensitivity is generally lower, so a dim marker that works on a conventional instrument may not work here. Run the panel on the actual instrument before committing a study to it.

**What does imaging flow cytometry do that conventional cytometry cannot?**

It captures an image of every event, so internalisation, colocalisation, nuclear translocation and morphology become measurable across a whole population rather than inferred from intensity or observed on a few cells.

**Is it slower than a conventional analyser?**

Yes, substantially, and that is the main practical cost alongside the data volume. For a panel that needs only intensity, a conventional analyser is faster, more sensitive and cheaper to run.

**Can it replace a microscope?**

No. Spatial resolution is far lower than a microscope's and there is no tissue context. It replaces the need to do microscopy on thousands of cells to get a population statistic.

**Do I still need the usual controls?**

All of them. Single stained controls for compensation, unstained cells, and a viability dye, exactly as on a conventional instrument. An image does not remove spectral spillover.

## Chemical testing labs

Analytical testing is bought by people who are not analysts, and the two recurring failures are ordering a test without naming a method and assuming that an accredited laboratory is accredited for the thing you asked. This page covers what to specify, what to check, and how to read what comes back.

1. **Name the analyte, the matrix and the method**: The same compound in water, in soil and in a formulated product is three different tests with three different preparations and three different detection limits. A request naming only the analyte will be quoted against whatever the laboratory assumes.
2. **Check the accreditation scope, not just the certificate**: Accreditation is granted for specific methods on specific matrices over specific ranges. The scope document is public and it is where you confirm the laboratory is accredited for your test, rather than for a test that sounds like it.
3. **State the detection limit you need**: A limit of quantitation below your decision threshold is the point of the exercise. A result reported as less than a limit that sits above your specification answers nothing, and this is discovered after the invoice more often than before.
4. **Agree sampling, or accept that the result is about the sample only**: A result describes what was tested. Where the sample was taken, by whom and how it was preserved decides what it represents, and a laboratory that did not sample cannot answer for that half of the measurement.
5. **Fix turnaround and the rush policy in writing**: Standard turnaround, the rush premium and what happens if a result fails quality control and needs a repeat. The repeat clause is the one nobody reads and the one that decides the real timeline.

**How do I check a chemical testing laboratory can do my test?**

Read its scope of accreditation, which names the methods, matrices and ranges it is accredited for. A laboratory holding accreditation is not thereby accredited for everything it can physically run.

**What has to be in a testing request?**

The analyte, the matrix, the method or the specification you are testing against, the detection limit you need, the number of samples and the turnaround. Without those, two quotations are not comparable.

**What does the measurement uncertainty on a report mean?**

The range within which the true value is expected to lie. It matters most when a result sits near a specification limit, because a result inside the limit with an uncertainty that straddles it is not a clear pass.

**Can a laboratory advise on what to test?**

Many will, and it is worth asking early. A conversation before the samples are taken frequently changes the method, the preservation and the number of samples, and it is much cheaper than repeating the exercise.

**Is an accredited result guaranteed for my sample?**

Only if your method and matrix are on the laboratory's schedule. Accreditation is granted method by method and matrix by matrix, so ask which published method will be used and at what reporting limit.

**Why is evidential testing different?**

Because chain of custody, sample integrity, analyst competence records and method defensibility are part of the deliverable. Forensic laboratories are accredited under schemes with additional requirements and write their reports differently.

## Indoor air quality testing

Air quality testing is frequently bought as a panel of measurements when what was needed was an investigation. Most indoor problems are ventilation, moisture or a specific source, and a broad screen without a hypothesis produces a table of numbers with no action attached. This page covers what to ask for, in what order, and how to read what comes back.

1. **Start with the complaint and a walkthrough**: Who is affected, where, when and since what change. A competent investigator walks the space, looks at the ventilation system and the moisture history, and forms a hypothesis before any sampling. Sampling without one is expensive guessing.
2. **Measure carbon dioxide as a ventilation proxy first**: It is cheap, continuous and it answers the commonest underlying question, which is whether the space is ventilated for its occupancy. Concentrations tracking occupancy and staying high indicate under-ventilation, and that explains a great many complaints on its own.
3. **Add temperature, humidity and particulate measurement**: Comfort parameters explain more complaints than contaminants do, and relative humidity drives both mold risk when high and irritation when low. Particulate measurement identifies combustion and outdoor ingress.
4. **Sample for specific contaminants only against a hypothesis**: Volatile organic compounds, formaldehyde, mold or carbon monoxide are targeted tests. Which method, over what duration, and where the sample is taken all follow from what you suspect, and a laboratory needs that to quote sensibly.
5. **Fix the reference you will read the result against before you get it**: Most indoor parameters have guidance values rather than legal limits, and occupational exposure limits are set for an industrial workforce rather than an office. Agree what a result will be compared against before it arrives, or it cannot be acted on.

**What should indoor air quality testing services measure first?**

Carbon dioxide, temperature, relative humidity and particulates, continuously and during occupancy. These are cheap, they identify the ventilation and comfort problems behind most complaints, and they tell you whether targeted contaminant sampling is even warranted.

**Are there legal limits for office air quality?**

Generally no, in the way people expect. Occupational exposure limits exist for workplaces with specific hazards and are set for an industrial context; office guidance values are advisory. Agree the comparison basis before commissioning the work.

**Should I test for mould?**

Only against a hypothesis. Air sampling for mold is difficult to interpret without an outdoor comparison and without knowing what normal is for that building. A moisture survey and a visual inspection usually answer the question better and cost less.

**How long should sampling run?**

Long enough to cover occupied and unoccupied periods, which usually means days rather than a single spot reading. A single reading at a quiet moment is the classic way to miss the problem entirely.

## Monoclonal antibody cost

A custom monoclonal antibody quotation is a stack of stages, and the same headline figure can cover very different scopes. The immunogen, the platform, how many clones are screened and how hard the screen is, and whether the result is delivered as a hybridoma or as a sequence are the four decisions that move the number. This page covers what each stage is for so that two quotations can be compared.

1. **The immunogen, which is cheap to buy and expensive to get wrong**: A peptide with carrier conjugation is inexpensive; a correctly folded recombinant protein or a membrane protein in a native conformation is a project in itself. A poor immunogen produces a campaign that finds nothing, which is the most expensive outcome available.
2. **The discovery platform**: Classical hybridoma work is well understood and takes months. Single B cell methods and display technologies are faster and cost more per campaign. The platform choice changes both the price and what kind of antibody you are likely to find.
3. **The screen, which is where projects succeed or fail**: Screening against the assay you will actually use costs more than screening on a generic binding assay, and it is the difference between a clone that works in your application and one that binds. Ask what the screening cascade is and insist your functional assay is in it.
4. **The number of clones taken forward**: Expanding, subcloning, freezing and characterising each candidate is per-clone work. A quotation for a campaign delivering one clone and one delivering five are different products, and the per-clone cost is the line to compare.
5. **Sequencing and recombinant conversion**: Delivering a hybridoma delivers a cell line that can drift or be lost; sequencing the variable regions and converting to a recombinant line makes the reagent permanent and reformattable. It costs more at the end and is almost always worth it for a reagent you will use for years.

**What drives monoclonal antibody production cost most?**

The immunogen difficulty, the platform, the screening cascade and the number of clones delivered. Between two quotations for the same target, the screening cascade and the clone count usually explain most of the difference.

**Why are two custom monoclonal antibody quotations so different?**

Because they rarely cover the same scope. Check the immunogen work, whether your functional assay is in the screen, how many clones are characterised and delivered, and whether sequencing and recombinant conversion are included.

**Should I pay for sequencing the clone?**

For a reagent you intend to use for years, almost always. A sequenced and recombinantly expressed antibody cannot drift, cannot be lost with a freezer failure and can be reformatted into any isotype or fragment later.

**Who owns the antibody?**

Whatever the contract says. Ownership of the immunogen, the hybridoma, the sequence, the right to publish and the right to have it made elsewhere are separate terms and defaults differ between providers. Settle them before work starts.

## IVD-grade antibodies

An antibody going into a diagnostic product is judged on things that barely matter for a research reagent: whether the supplier will tell you before they change anything, whether they will reserve a lot, and whether they can produce the documentation a device manufacturer's quality system requires. Affinity and specificity are necessary and they are not what separates the grades.

1. **Establish the grade and what it commits the supplier to**: Research use only material carries no commitment to consistency or notification. A reagent sold for diagnostic manufacture should come with a specification, a certificate per lot against it, and a stated change control process. Ask what the grade actually obliges the supplier to do.
2. **Get change notification in the contract**: A supplier changing a purification step, a host cell line or a buffer without telling you can invalidate a validated assay silently. Notification before change, with enough lead time to bridge, is the single most valuable term in the agreement.
3. **Reserve a lot and bridge between lots**: Reserve enough of one lot to cover production and validation, and run a bridging study when moving lots, comparing performance in the actual assay rather than on a binding measurement. Lot change is the commonest cause of a diagnostic assay drifting.
4. **Collect the documentation your quality system needs**: Specification, certificate of analysis per lot, animal origin and source material statements, a supplier quality agreement and audit rights. Supplier controls are a regulated requirement for a device manufacturer, not paperwork for its own sake.
5. **Assess supply risk as seriously as performance**: A single source reagent with no second supplier is a single point of failure for the product. Ask about capacity, about how long the supplier commits to producing it, and what notice of discontinuation you would receive.

**What makes an ivd antibody different from a research antibody?**

Commitments rather than chemistry: a written specification, a certificate per lot, change notification, lot reservation and documentation your quality system can use. The same clone can be sold at both grades with very different obligations attached.

**Can I use a research use only antibody in a diagnostic product?**

The labeling restricts the supplier's claim, and the obligation to qualify what goes into a device is the manufacturer's. In practice a reagent with no specification, no change control and no lot certificate is very difficult to qualify and to keep qualified.

**Why is change notification so important?**

Because an undisclosed process change can shift assay performance without anything visibly failing. Notification with lead time is what lets you bridge lots deliberately instead of discovering a shift in production.

**What documentation should a supplier provide?**

A product specification, a certificate of analysis per lot against it, source and animal origin statements, and a quality agreement covering change notification and audit rights. Ask for a sample set before committing.

## Antibody sequencing

Sequencing an antibody turns a perishable cell line or a dwindling vial into a permanent, reproducible reagent, and it is the single most valuable thing a laboratory can do to an antibody it depends on. It is also routinely left until the hybridoma has already been lost, at which point the work has to be done from purified protein. This page covers both routes and the proof that matters.

1. **From the cell, which is far easier, while you still have it**: If the hybridoma or the producing cell line is alive, the variable region sequences are recovered from its messenger RNA by amplification and sequencing. It is cheap, fast and unambiguous. Do it while the line is healthy rather than after it is not.
2. **From purified protein, when the cell is gone**: De novo sequencing by mass spectrometry reconstructs the sequence from overlapping peptides. It needs a few hundred micrograms of purified antibody, several proteases to produce overlapping coverage, and it is a considerably larger analytical exercise.
3. **Insist on full coverage of the complementarity determining regions**: The binding regions are where errors matter and where coverage is hardest, because they are the most variable. A report with gaps in those loops has not delivered a usable sequence, whatever the overall coverage percentage says.
4. **Watch the isobaric residues**: Leucine and isoleucine have identical mass and cannot be distinguished by ordinary mass spectrometry, and there are similar traps. A good provider states how it resolved them, or states that it did not, and an unresolved position has to be tested by expression.
5. **Prove the sequence by expressing it**: The only real validation is to express the recombinant antibody from the reported sequence and show it binds like the original, in the assay you care about. A sequence that has not been expressed is a hypothesis, and the expression test should be part of the project.

**Can an antibody be sequenced from purified protein alone?**

Yes, by de novo mass spectrometry using several proteases to generate overlapping peptides. It needs a few hundred micrograms of pure antibody and it is harder and more expensive than sequencing from the cell that makes it.

**How much antibody do protein sequencing services need?**

Typically a few hundred micrograms of purified, homogeneous material. Contaminating immunoglobulin from serum or a carrier protein in the formulation complicates the analysis considerably, so supply carrier-free purified material.

**How accurate is de novo antibody sequencing?**

Good providers achieve high coverage, and the difficult part is the binding loops and residues with identical mass. Expressing the sequence and showing the recombinant binds like the original is the only proof that settles it.

**Why sequence a hybridoma antibody at all?**

To make it permanent. A sequence can be expressed indefinitely, reformatted into any isotype or fragment, produced anywhere and shared, while a hybridoma drifts, can be lost and exists in one freezer.

**Is hybridoma sequencing easier than sequencing the protein?**

Far easier, while the line is alive. If the hybridoma or the producing cell line is healthy, the variable region sequences are recovered from its messenger RNA by amplification and sequencing, which is cheap, fast and unambiguous. Sequencing from purified protein is the route for when the cell is gone, so do it while the line is still growing.

## Capillary electrophoresis

Capillary electrophoresis separates by movement in an electric field rather than by partition onto a stationary phase, which makes it excellent at things chromatography finds hard: charge variants, small ions, and nucleic acid fragment sizing. It is also notorious for migration time drift, and almost all of that is capillary surface conditioning. This page covers the modes, the specification and the conditioning.

1. **Understand the separation principle**: Analytes move through a buffer-filled capillary under a high electric field, separating on the ratio of charge to hydrodynamic size, with bulk flow from the charged capillary wall carrying everything along. There is no stationary phase, which is why the capillary surface matters so much.
2. **Pick the mode from the analyte**: Zone electrophoresis in free solution for small ions and charged molecules. Gel-filled capillaries for sizing nucleic acids and for protein sizing as an alternative to a gel. Isoelectric focusing for protein charge variants, which is one of the technique's strongest applications.
3. **Specify capillary dimensions and the detector together**: Narrow capillaries dissipate heat better and allow higher fields, at the cost of a shorter detection path length and so less sensitivity with absorbance detection. Fifty to seventy five micrometres internal diameter covers most work, and laser induced fluorescence is what rescues sensitivity where it is needed.
4. **Condition the capillary, and do it the same way every time**: Migration time drift is almost always the capillary wall changing as sample components adsorb. A documented rinse sequence between runs, and a longer conditioning at the start of a sequence, is what turns an irreproducible method into a reproducible one.
5. **Control temperature and injection precisely**: Both migration time and peak area depend on temperature and on the injection, which is by pressure or by voltage rather than by a fixed loop. An internal standard is close to mandatory for quantitation.

**What does a capillary electrophoresis system do better than HPLC?**

Charge variant separations, small ion analysis and nucleic acid fragment sizing, with tiny sample volumes and very high efficiency. It is not a general replacement for chromatography and it is much better at those particular jobs.

**Why do my migration times drift?**

The capillary surface is changing as sample components adsorb, which alters the bulk flow. A consistent rinse and conditioning sequence between runs, and a coated capillary for sticky samples, fix most of it.

**Is capillary electrophoresis quantitative?**

Yes, with care. Injection is by pressure or voltage rather than a fixed loop and peak area depends on migration velocity, so an internal standard and corrected peak areas are the normal practice rather than an optional refinement.

**When should I use a coated capillary?**

When proteins or other sticky analytes adsorb to the wall, which shows as tailing and drifting migration. Coatings suppress the bulk flow and the adsorption together, and they are consumables with a limited life.

## Headspace analysis

Headspace sampling measures the vapour above a sample instead of injecting the sample itself, which keeps involatile matrix out of the chromatograph entirely. That is its whole advantage and the source of its one persistent trap: what reaches the vapour depends on the matrix, so a calibration made in a different matrix is not a calibration. This page covers the modes and the method.

1. **Static headspace for a known volatile at a workable concentration**: The vial is equilibrated at temperature and a fixed volume of the vapour is injected. It is simple, robust and reproducible, and it is the default for residual solvents and for alcohol determinations.
2. **Dynamic purge and trap for trace work**: Inert gas is bubbled through the sample and the volatiles are collected on a trap and then desorbed, which concentrates them by orders of magnitude. It is what environmental volatile methods use and it is more complex and more prone to carryover.
3. **Fix equilibration temperature and time, and prove the equilibrium**: Both control how much analyte reaches the vapour. Run a time series at your chosen temperature and confirm the response has plateaued, because a method sampling before equilibrium is exquisitely sensitive to timing variation.
4. **Match the calibration matrix or use standard addition**: Partition into the headspace depends on the sample matrix, so a standard curve prepared in water will not describe a solvent, an oil or a polymer. Matrix-matched standards or standard addition are the two honest routes, and skipping this is the commonest error in headspace work.
5. **Control vial fill, seal and carryover**: The ratio of sample to headspace volume is a method parameter and must be constant. Septa that leak or core cause low results, and the transfer line and loop need to be hot enough and purged enough that a strong sample does not appear in the next one.

**Static or dynamic headspace analysis?**

Static for known volatiles at concentrations it can reach, which covers residual solvent testing and most quality control work. Dynamic purge and trap where trace sensitivity is required, at the cost of complexity and a greater carryover risk.

**Why do my results change when the matrix changes?**

Because partition into the vapour depends on the matrix. A calibration in one matrix does not describe another, which is why matrix-matched standards or standard addition are required rather than recommended.

**How do I choose the equilibration temperature?**

High enough to get the analyte into the vapour in a reasonable time, low enough not to degrade the sample or over-pressurise the vial. Then prove the response has plateaued at that temperature with a time series.

**What causes carryover in a headspace analyzer?**

A cool transfer line, an inadequate loop purge, or a strongly retained analyte on a trap in a dynamic system. Run a blank after your highest standard as routine, because carryover here is easy to miss and it inflates every following result.

## Optical emission

Optical emission spectrometry excites atoms and reads the light they emit, and the source determines what kind of laboratory the instrument belongs in. A plasma source measures elements in solution across most of the periodic table; a spark source measures a solid metal directly on the shop floor. Choosing between them is choosing which problem you have. This page covers both and the calibration that makes either defensible.

1. **Decide whether your sample is a solution or a solid metal**: An inductively coupled plasma instrument nebulises a solution, so everything must be digested or dissolved first. A spark instrument burns a prepared surface of a solid metal directly and gives an alloy composition in under a minute. They are not alternatives for the same sample.
2. **Match the detection limits to your specification**: Plasma emission covers most elements at low concentrations and is displaced by mass spectrometric detection when limits fall to the trace and ultratrace region. Write the limits you need per element before comparing instruments, because performance varies by element far more than headline figures suggest.
3. **Plan the digestion, which is most of the work**: For plasma work the sample preparation dominates the error budget and the labor. Microwave digestion with the right acid mixture, in clean vessels, is a laboratory capability in itself and is where contamination enters.
4. **Calibrate against matrix-matched certified standards**: Emission intensity depends on the matrix, so standards must resemble the sample. Certified reference materials give traceability, and an internal standard corrects for nebulisation and plasma variation run to run.
5. **Watch for spectral interference and correct it deliberately**: Emission lines overlap, and an interfering element can inflate a result substantially. Choose analytical lines with the interferences in mind, apply the correction the software offers knowingly, and verify against a certified material containing the interferent.

**ICP or spark optical emission spectrometer?**

Plasma for solutions across most elements, after digestion, with good detection limits. Spark for direct analysis of solid metal samples, which is fast and is what a foundry or a metal fabricator needs on site.

**When is mass spectrometric detection needed instead?**

When detection limits fall into the trace and ultratrace region, or when isotope information is needed. Plasma emission is cheaper to buy and run and is sufficient for a great deal of routine elemental work.

**Why do results differ from another laboratory?**

Usually digestion or calibration matrix rather than the instrument. Incomplete dissolution, a different acid mixture, or standards in a matrix unlike the sample all shift results, and all three are invisible in the final report.

**What consumables and services does it need?**

Argon in quantity for a plasma instrument, torch components, nebuliser and spray chamber, plus an exhaust. Argon supply is a significant recurring cost and belongs in the purchase decision rather than in the first year's budget surprise.

## Service contracts

Service contracts are usually bought as a bundle at instrument purchase and renewed without review, which is how a laboratory ends up paying a full contract on a centrifuge and nothing on the instrument whose failure stops everything. The decision is per instrument and it turns on one question: what does a week of downtime cost. This page covers how to make it.

1. **Start from what a failure actually costs**: An instrument on the critical path whose failure stops sample processing justifies a contract; a bench item with a backup does not. Write the consequence of a week of downtime next to each instrument, and the list sorts itself.
2. **Read what response time means**: Response is frequently defined as a telephone call rather than an engineer on site, and the clock may run on business hours only. Ask for the definition, the on-site target and the historical achievement, and get the achievement rather than the target in writing if you can.
3. **Check whether parts are included, and which parts**: Contracts commonly exclude consumables and sometimes exclude the expensive items such as detectors, lamps and pumps. A contract excluding the part most likely to fail is close to worthless, and the exclusions are where the margin is.
4. **Ask about a loan instrument**: For an instrument on the critical path, a loan while yours is repaired is worth more than a faster response. Not every vendor offers one and the ones that do rarely lead with it.
5. **Decide manufacturer against third party per instrument**: Manufacturer service has the parts, the software and the qualification documentation. A good third party is cheaper and may respond faster locally. For regulated instruments the qualification documentation frequently decides it; for a shaker it does not.

**Which lab instruments need a service contract?**

Those whose failure stops work and whose repair needs the manufacturer: biosafety cabinets, ultra low freezers, mass spectrometers, cytometers and anything with optics or a detector. Simple mechanical items are usually cheaper repaired on failure.

**What does response time mean in lab instrument service contracts?**

Read the definition. It is often a telephone response within business hours rather than an engineer on site, and the two can differ by a week. Ask for the on-site target and what it has historically been.

**Is third party service acceptable?**

Frequently yes, and cheaper. The questions are parts access, whether they can supply qualification documentation for regulated instruments, and whether using them affects any remaining warranty.

**Should preventive maintenance be included?**

For instruments where it genuinely prevents failure, yes, and make sure it includes the qualification you need afterwards. A preventive visit that leaves an instrument needing a separate qualification has done half a job.

## Laboratory reagents

Reagent grade is bought by habit and it decides results. An analytical grade certifies impurities to a stated limit; a technical grade certifies very little; a molecular biology grade certifies the absence of the specific contaminants that ruin enzymatic work. Buying above what the method needs wastes money and buying below it produces failures nobody attributes correctly. This page covers grade, lot and shelf.

1. **Match the grade to what the method is sensitive to**: Analytical grade for quantitative chemistry, chromatography grade where trace impurities raise a baseline, molecular biology grade where nuclease or nucleic acid contamination matters, and technical grade for cleaning and for bulk uses where purity is irrelevant.
2. **Read the certificate of analysis rather than the label**: The certificate names the impurities tested and their limits for that lot. Two suppliers' analytical grades can certify different impurity sets, and the one that matters to your method may be on only one of them.
3. **Record the lot number with the result, not with the invoice**: When an assay drifts, reagent lot is the first variable to check and it is usually unrecorded. Writing the lot into the method record costs nothing and turns a week of investigation into one query.
4. **Date every container on receipt and on opening**: Shelf life runs from manufacture and many reagents degrade faster once opened. A bottle with no opening date is a bottle nobody can judge, and it will be used long after it should have been discarded.
5. **Keep the safety data sheet and label every decant**: Every hazardous chemical needs its sheet available and every container other than one in immediate use needs its identity and hazards on it. An unlabelled bottle is the single most common laboratory inspection finding.

**What grade of laboratory reagents do I need?**

The grade the method is sensitive to. Analytical for quantitative chemistry, chromatography grade for low wavelength ultraviolet and mass spectrometric detection, molecular biology grade for enzymatic work, technical for cleaning and bulk use.

**Does reagent lot really matter?**

Often more than the supplier does. Lot to lot variation in serum, enzymes, antibodies and even some salts is real, and an assay that drifts after a reorder is a lot change until proven otherwise. Record the lot with every result.

**How long do reagents last once opened?**

It depends entirely on the reagent and on how it is stored, which is why the opening date on the bottle matters more than a generic rule. Hygroscopic solids, reducing agents and anything light sensitive degrade fastest.

**Should we standardise on one supplier?**

For convenience and price, largely yes, with a second source for anything whose absence stops work. What should not vary within a study is the supplier and the lot of the reagents the result depends on.

**Are two buffers at the same pH interchangeable?**

No. Phosphate precipitates with divalent cations and inhibits some enzymes; tris shifts pH with temperature. Choose the species for the chemistry, titrate at the temperature of use and record the recipe rather than the name.

**Why is the calcium and magnesium free formulation used for washing?**

Because those ions support the adhesion molecules holding cells together and to the surface. Omitting them loosens cells before dissociation; including them keeps tissue intact. Using the wrong one is an invisible protocol error.

## Trial support services

Trial support covers a wide range of functions that are bought together and behave differently: monitoring, data management, central laboratory work, drug supply, statistics and medical writing. Outsourcing transfers the work and never the responsibility, which is the sentence most sponsors learn late. This page covers what to outsource, how to keep oversight and what to fix in the contract.

1. **Separate the functions before pricing them**: Monitoring, data management, central laboratory, drug supply, statistics and medical writing have different markets and different specialists. Bundling them with one provider buys convenience; separating them buys the best of each and costs coordination effort.
2. **Keep oversight, because the obligation stays with the sponsor**: Regulatory responsibility for trial conduct remains with the sponsor whoever performs the work, so the oversight plan is a deliverable in its own right: what you review, how often, and what triggers escalation.
3. **See the data as the study runs, in an analysable form**: Agree interim transfers at defined intervals in a format you can actually analyze. A sponsor seeing its own data monthly finds design and site problems while they are still fixable; one waiting for the final transfer finds them afterwards.
4. **Name the people, not just the company**: The project manager, the lead monitor and the data manager determine how the study goes. Name them in the contract, agree notice on replacement, and meet them before signing.
5. **Settle data, documents and site relationships at the end**: What is returned at close-out, in what format, who holds the trial master file, how long records are retained and whether you may approach the sites directly afterwards. These are contract terms and they are difficult to add later.

**What do clinical trial support services usually cover?**

Monitoring, data management, central laboratory analysis, investigational product supply, biostatistics, medical writing and regulatory submission support. They are frequently bought as a bundle and are distinct markets with distinct specialists.

**Does outsourcing transfer responsibility?**

No. The sponsor retains regulatory responsibility for trial conduct however much is delegated, which is why a documented oversight plan is expected and why delegation has to be recorded rather than assumed.

**Should everything go to one provider?**

A single provider reduces coordination and gives you one relationship to manage and one point of failure. Splitting functions gets specialists and costs integration effort. Study size and internal capacity usually decide which is right.

**What is the most commonly missed contract term?**

What comes back at the end: the data in an analysable form, the trial master file, the site contacts and the right to approach them. Close-out is negotiated at the start or it is not negotiated at all.

**What is the difference between a monitor and a medical monitor?**

A monitor verifies consent, eligibility, protocol compliance and source data at sites. A medical monitor is a physician advising on eligibility questions and safety events. Contracts that conflate them leave one role unstaffed.

**Are clinical trial payments to participants allowed?**

For time and inconvenience, with the arrangement reviewed by the ethics committee. It may not be an inducement, and poor payment logistics are a real and avoidable cause of dropout.

## Sample tracking

Tracking is a physical problem before it is a software one. An identifier that fades, a transfer nobody scanned and an inventory nobody has audited will defeat any system, however good. This page covers the three decisions that decide whether a tracking system still describes reality a year after it went live.

1. **Choose an identifier that survives what the sample survives**: Labels lift in solvent, frost and liquid nitrogen, and a printed adhesive label on a cryovial is the commonest failure. Two dimensional codes molded or laser marked into the vial base read reliably under frost and cannot come off, and they are worth the higher unit cost from the first study.
2. **Make the identifier meaningless**: An identifier encoding the study, the date and the subject looks helpful and becomes wrong the moment any of those change. Use an opaque identifier and hold the meaning as data, so a re-assignment is an edit rather than a relabelling exercise.
3. **Decide the scan points and make them the only way to move a sample**: Receipt, aliquot, storage, retrieval, transfer and disposal. A workflow where the position can be typed rather than scanned is a workflow that will drift, and the drift is invisible until somebody cannot find something.
4. **Record position to the slot, not to the freezer**: Freezer, rack, box, position, with a visible box map. Searching with the door open warms everything around the target, and a system that stops at the freezer number has not removed the problem it was bought for.
5. **Reconcile physically, on a schedule**: Count a sample of boxes against the record and record the discrepancy rate. Every collection has one, and knowing it is the difference between an inventory you can act on and one you hope is right. Audit after every move, freezer failure and change of staff.

**What identifier should a lab sample tracking system use?**

An opaque, meaningless identifier carried by a two dimensional code that survives the sample's storage conditions. Codes on the vial base read under frost where side labels do not, and meaning belongs in the data rather than in the string.

**How detailed does position tracking need to be?**

To the slot in the box. Anything coarser means searching with the door open, which harms neighbouring samples, and it makes a physical audit impractical to complete.

**How often should the inventory be audited?**

On a schedule, and after every move, freezer failure or change of staff. Those three events cause most of the drift between a record and a freezer, and an audit right after one of them is worth several routine ones.

**Can a spreadsheet do this?**

Until two people edit it at once, or somebody needs to know where a vial was six months ago, or a consent has to be traced through aliquots. At that point the spreadsheet is the risk rather than the saving.

## Epitope tag antibodies

Tagging is the way out of the antibody problem. Instead of finding a reagent that recognizes your protein, you attach a short sequence that an extremely well characterised reagent already recognizes, and every experiment afterwards uses the same validated antibody. The decisions that remain are which tag, where to put it and how many copies, and each of those can break the protein rather than the detection.

1. **Understand what the tag buys**: One reagent, validated once, works on every tagged construct in the laboratory, in every application, with a known negative control in the untagged parent. That reproducibility is worth more than almost any improvement in an antibody against the native protein.
2. **Choose between the common tags on practical grounds**: The short tags differ mainly in length, charge and which reagents are best characterised for which application. Where two tags are needed on different proteins in one experiment, choose ones whose antibodies are raised in different hosts so the secondaries do not collide.
3. **Decide the terminus from the protein's biology**: A tag at a terminus that carries a signal sequence, a localisation signal or a processing site will disrupt it. Check both termini against what is known about the protein, and if neither is safe, an internal tag in a flexible loop is the alternative.
4. **Use multiple copies when signal is the problem, and expect a cost**: Repeating the tag increases the signal proportionally and increases the chance of interfering with folding, trafficking or interactions. Start with one copy and add only if detection genuinely fails.
5. **Prove the tagged protein still behaves**: Localisation, molecular weight, an activity assay and, where possible, rescue of the untagged protein's function. A tagged construct that has not been shown to behave like the original is a new protein you are now studying.

**Why use a V5 tag antibody instead of one against my protein?**

Because the reagent is already characterised and the same one works for every tagged construct. You also get a clean negative control for free in the untagged parent line, which is the hardest control to obtain otherwise.

**Which tag should I choose, an HA tag antibody or a Myc tag antibody?**

Any of the well established short tags will do for detection. Choose on practical grounds: antibody host species where you need to multiplex, whether the tag's charge or length is likely to matter, and which reagents your laboratory already has validated.

**Amino or carboxy terminus for a FLAG tag antibody?**

Whichever does not disrupt a signal sequence, a localisation signal, a processing site or a domain boundary. If both are problematic, an internal site in a flexible loop is worth trying, though it needs more validation.

**Does tagging change the protein, and will a Strep tag II antibody still bind?**

Sometimes, and that is why the control experiments matter. Show that the tagged protein localises correctly, runs at the expected size and retains function before trusting any result from it.

**Why does my tag antibody give a band in untransfected cells?**

Because several common tags come from real proteins, and the antibody sees the endogenous version too. Run an untransfected control in every experiment so the tagged band can be told apart from it.

## Biotinylated antibodies

Biotin and streptavidin bind more tightly than almost any other non-covalent pair, which makes the system the standard way to amplify a weak signal. It also has one specific failure mode that produces confident, wrong results: tissues that contain their own biotin light up whether or not your antibody bound anything. This page covers both halves.

1. **Understand where the amplification comes from**: Several biotins are attached to each antibody and each can bind a streptavidin conjugate carrying several reporter molecules, so the signal multiplies at two stages. That is why the system rescues weak targets that direct conjugates cannot detect.
2. **Block endogenous biotin in tissues that carry it**: Liver, kidney, brain and several other tissues contain substantial free biotin. Without a blocking step using avidin followed by free biotin, those tissues stain regardless of the primary antibody, and the result looks like real staining.
3. **Run the primary-free control in every tissue**: The control that exposes endogenous biotin is the one with the detection system and no primary antibody. In a tissue you have not used before, that control is not optional and it costs one slide.
4. **Watch the biotin in the sample and the medium**: Culture media and serum supplements contain biotin, and samples from supplemented subjects can carry enough to interfere with an immunoassay. This is a known and documented source of erroneous immunoassay results and it is worth knowing about before an unexplained value appears.
5. **Titrate both halves, not just the antibody**: The biotinylated primary and the streptavidin conjugate both need titrating, and an excess of either raises background. Amplification multiplies the background as faithfully as the signal.

**When should I use a biotinylated antibody?**

When signal is the limitation and a directly conjugated reagent is not sensitive enough. The biotin and streptavidin system amplifies at two stages and is the standard route for low abundance targets in tissue.

**What is endogenous biotin and why does it matter?**

Biotin present naturally in tissues, particularly liver, kidney and brain. It binds the streptavidin conjugate directly, so those tissues stain with no primary antibody at all. Blocking with avidin then free biotin prevents it.

**Does a biotin labeled antibody work in flow cytometry?**

Yes, with a streptavidin conjugate as the second step, and it is a common way to build a panel where a needed clone is not available in the fluorophore you want. It adds a wash step and requires its own titration.

**Can biotin in a sample interfere with an immunoassay?**

It can. High biotin intake raises circulating biotin enough to interfere with assays built on this system, which is a documented cause of erroneous clinical results and a reason to ask what a sample donor was taking.

**When do I need an anti biotin antibody instead?**

When biotin itself is the target: detecting a biotinylated probe on a blot where streptavidin would bind endogenous biotin too, or measuring how much biotin is on a molecule. It is the opposite reagent from a biotinylated antibody, and confusing the two is a common ordering mistake because the names look alike.

## LIMS by industry

Vendors sell one product and configure it, and the configuration is where the industry lives. Three of the commonest configurations behave so differently that they are effectively different systems: water testing, stability programs, and manufacturing quality control. This page sets out what each one actually demands, so a requirement can be written before a demonstration is booked.

1. **Water testing: methods, holding times and the deliverable**: The clock starts at collection, every analyte has its own holding time and preservation, quality control batching follows the method, and the electronic deliverable has to arrive in the format the receiving agency accepts. A system that stores a collection date without acting on it has not solved the problem.
2. **Stability: a schedule that pulls samples for years**: A stability program is a calendar. The system holds the protocol, generates the pull points years ahead, tracks which chamber each sample is in, and warns before a pull is missed. A missed time point cannot be recovered, which is why scheduling rather than testing is the core function.
3. **Manufacturing: results tied to batches and specifications**: Every result belongs to a batch, a specification decides pass or fail, and a certificate of analysis is issued against it. The system has to talk to the manufacturing execution or planning system and to hold an out of specification investigation as a workflow rather than a note.
4. **Ask which configuration the vendor actually runs elsewhere**: All three exist as configurations of the same products, and a vendor with ten water laboratories has solved the problems your water laboratory will hit. Ask for a reference of your industry and your size, and ask that reference how long implementation took.
5. **Write the requirement before the demonstration**: A demonstration shapes itself around the demonstrator's strengths. A written requirement naming your methods, your holding times or your pull schedule, and your reporting obligations, is what makes two demonstrations comparable.

**What does a water or wastewater LIMS need that a general system does not?**

Holding times computed from collection and enforced, preservation and quality control batching driven by the method, chain of custody as data, and electronic deliverables in the formats regulators accept.

**What is different about a stability lims?**

It is a scheduling system first. It has to generate pull points years ahead from the protocol, track which chamber holds each sample, and warn before a pull is missed, because a missed stability time point cannot be recovered.

**How does lims manufacturing differ from a testing laboratory system?**

Everything is attached to a batch and judged against a specification, certificates of analysis are issued, out of specification investigations are a workflow, and the system usually has to exchange data with manufacturing planning.

**Can one product serve all three?**

Most major products can be configured for any of them. What differs is how much configuration, and whether the vendor has done it before for a laboratory like yours, which is the question worth asking rather than whether it is possible.

## Digital pathology

A slide scanner is a microscope that must work unattended on hundreds of slides, which changes what matters about it. Optical quality is necessary and rarely the limitation; focus strategy, color consistency, throughput and what happens to the images are what decide whether a digital pathology program succeeds. This page covers those.

1. **Decide the magnification you actually need**: Scanning at twenty times is faster, produces smaller files and suffices for most morphology. Forty times is needed where nuclear detail or small organisms matter, and it roughly quadruples both scan time and file size. Decide per application rather than defaulting to the higher setting.
2. **Interrogate the focus strategy, because it is where scanners fail**: A whole slide is not flat and tissue is not one thickness. How the scanner builds its focus map, how many focus points it takes and whether it can capture multiple planes decides whether thick or folded sections come back usable. Ask for your worst slides to be scanned as the demonstration.
3. **Insist on color consistency and calibration**: Two scanners, or one scanner over a year, drift in color, and stain intensity is part of how a slide is read. A calibration routine and a reference slide are what make images comparable between instruments and over time.
4. **Match throughput and capacity to the real workload**: Slides per hour at your chosen magnification, with the loader capacity and whether slides can be added while it runs. A scanner that must be emptied before reloading sets the daily output rather than the scan speed.
5. **Plan the storage and the viewing before the scanner arrives**: Images run to hundreds of megabytes or a gigabyte each at higher magnification, so a busy scanner produces terabytes a year. Decide where the images live, who can view them, how they are backed up and how long they are kept, before the first batch.

**What magnification should a microscope slide scanner use?**

Twenty times for routine morphology, forty times where nuclear detail or small structures matter. Forty times roughly quadruples scan time and file size, so it is a per-application decision rather than a default.

**Why do some scanned slides come back out of focus?**

The focus map. A whole slide is not flat and tissue thickness varies, so the number and placement of focus points and the ability to capture several planes decide whether awkward sections are usable. Test with your worst slides.

**How much storage does digital pathology need?**

Hundreds of megabytes to about a gigabyte per slide depending on magnification and tissue area, so a busy service produces terabytes a year. Storage, backup and a retention policy belong in the business case, not in the following year's budget.

**Can scanned images be used for diagnosis?**

That depends on the regulatory status of the specific system and on local validation requirements, and it is a different question from whether the images look good. Establish the intended use before buying, because a research scanner and a diagnostic one are different products.

## Oligo manufacturing

An oligonucleotide made at research scale and one made at production scale are the same chemistry with completely different economics and controls. Coupling efficiency compounds over length, purification dominates the cost as batches grow, and the analytics and change control become the contract. This page covers what changes and what a supply agreement has to say.

1. **Coupling efficiency compounds, so length is the cost driver**: Each addition is a chemical step at slightly less than complete efficiency, and the full length yield falls geometrically with length. That is why a long oligonucleotide costs disproportionately more and why purification matters more the longer it gets.
2. **Purification, not synthesis, sets the cost at scale**: Desalted material is cheap; cartridge, high performance liquid chromatography and gel purification each add a step and a yield loss. Specify the purification the application actually needs, because buying gel purity for a screening primer is a large recurring waste.
3. **Modifications multiply everything**: Every dye, quencher, phosphorothioate linkage or modified base is a separate reagent with its own coupling behavior and its own cost, and some combinations constrain the purification route. List every modification when asking for a quotation.
4. **Fix the release testing in the specification**: Mass spectrometry confirming identity, a purity measurement by a stated method, and concentration by absorbance are the usual release package. A certificate that reports concentration alone is a shipping note rather than a release.
5. **Write change control into the supply agreement**: A supplier changing a synthesis site, a reagent source or a purification route can shift impurity profiles in ways an assay will feel. Notification with lead time, and lot reservation, are the terms worth more than a small price difference.

**Why does a longer oligonucleotide cost so much more?**

Because coupling efficiency compounds. Each addition is slightly less than complete, so the proportion of full length product falls geometrically with length, and more material must be made and purified to deliver the same amount.

**What purification do I actually need?**

Desalted for routine primers, cartridge or chromatographic purification for probes and anything where truncated sequences interfere, and the highest grades for therapeutic or diagnostic applications. Buying above the need is a large recurring cost.

**What should oligonucleotide analysis on a certificate show?**

Identity by mass spectrometry against the expected mass, purity by a named method with the value, and concentration with the method used. Those three make the material traceable; a concentration alone does not.

**How do I secure supply from an oligonucleotide manufacturer?**

A supply agreement with change notification, lot reservation, an agreed specification and a stated lead time, plus a qualified second source where the volume justifies the second validation.

## Clinical centrifuges

A centrifuge is chosen by its rotor rather than by its top speed, and clinical work adds two constraints a research laboratory can ignore: the tubes are standard blood collection tubes with gels and separators that need a particular geometry, and the samples are potentially infectious. This page covers the rotor decision and the containment one.

1. **Specify in relative centrifugal force, never in revolutions**: The force a sample experiences depends on the rotor radius as well as the speed, so the same revolutions per minute in two rotors give different separations. Protocols are written in relative centrifugal force, and a laboratory converting on the fly will eventually get it wrong.
2. **Choose swing-out for separated layers, fixed angle for pellets**: A swing-out rotor gives a horizontal, flat interface, which is what a serum or plasma separator tube needs and what makes a clean draw possible. A fixed angle rotor packs a tight pellet against the tube wall and is faster and more compact.
3. **Match the buckets and adaptors to the exact tubes**: Blood collection tubes vary in length and diameter between manufacturers, and adaptors are rotor-specific. A rotor that does not take your tubes without a wobble will break them, and tube breakage in a clinical laboratory is a containment event.
4. **Decide refrigeration from what you spin**: Cell separations and anything where analyte stability depends on temperature need refrigeration. For routine serum separation at ambient it is a cost without a benefit, and refrigerated instruments are larger and noisier.
5. **Take containment seriously for clinical samples**: Sealed buckets or a sealed rotor that can be opened inside a biosafety cabinet are what turn a broken tube from an aerosol release into a contained spill. For potentially infectious material this is the specification that matters most.

**Swing-out or fixed angle centrifuge?**

Swing-out for separated layers and gel separator tubes, because the interface forms flat and a clean draw is possible. Fixed angle for pellets, where it packs tighter, spins faster and takes less bench space.

**Why do protocols specify RCF rather than rpm?**

Because the force depends on rotor radius as well as speed. The same revolutions per minute in a large and a small rotor produce different separations, so the force is the transferable number and the speed is not.

**What containment does a clinical centrifuge need?**

Sealed buckets or a sealed rotor that can be carried into a biosafety cabinet and opened there. A tube breaking in an unsealed rotor releases aerosol into the room, which is the event the sealing exists to prevent.

**How often should a centrifuge be serviced?**

Annually as a minimum, with rotor inspection for corrosion and fatigue and a check of the lid interlock and the imbalance detection. Rotors have documented service lives and a failed rotor at speed is a serious hazard.

**Can a general bench instrument replace a microhematocrit centrifuge?**

Usually not. The measurement needs a high force for a short time in a capillary geometry, read against a scale made for it. A dedicated instrument is small, inexpensive and the right tool for that one job.

## Cytokine ELISA kits

Cytokines sit at picogram concentrations, are released from cells during clotting and handling, and are often partly bound to soluble receptors and binding proteins that different kits see differently. That combination makes the pre-analytical protocol and the kit's antibody pair more important than almost anything else about the assay. This page covers both.

1. **Match the standard range to what you expect**: Most cytokines circulate at very low picogram levels in health and rise sharply in disease, so a kit whose range brackets healthy values may saturate in patients and vice versa. Pick the range that puts your expected values in the middle of the curve where precision is best.
2. **Establish what the kit detects**: Free cytokine, total including that bound to soluble receptors, or something in between depends on the antibody pair and on whether a dissociation step is included. Two kits measuring the same cytokine can legitimately give values that differ several-fold for this reason.
3. **Fix the pre-analytical protocol and follow it exactly**: Tube type, time to centrifugation, temperature and freeze-thaw all move cytokine results, and several are released from leucocytes and platelets during clotting. Serum and plasma are not interchangeable and the anticoagulant matters.
4. **Qualify the kit in your own matrix**: Spike a known amount into your actual sample type and recover it, and check that a serial dilution back-calculates consistently. A kit validated in serum used on culture supernatant or lysate is unvalidated until you show otherwise.
5. **Decide single analyte or multiplex deliberately**: A multiplex panel measures many cytokines from one small volume and brings cross-reactivity and a compromise buffer with it. For one analyte that matters, a dedicated kit is usually more accurate; for a broad survey a panel is the only practical route.

**Why do two human il-6 elisa kits give different values?**

Different antibody pairs see different forms, including cytokine bound to soluble receptor, and different standards are calibrated differently. Values from two kits are generally not interchangeable, which is why one kit and one lot should run a whole study.

**Serum or plasma for cytokine measurement?**

Whichever the kit was validated in, applied consistently. They are not interchangeable: several cytokines are released from cells during clotting, so serum values can be substantially higher than plasma values from the same donor.

**How do I qualify elisa kits for cytokines in my matrix?**

Spike recovery and linearity of dilution in the actual sample type. Recovery close to the amount added and parallel dilution are what show the matrix is not suppressing or enhancing the signal.

**Is a multiplex panel or cytokine antibody array as good as a single kit?**

For a survey, it is the only practical route and it is good enough. For one analyte driving a conclusion, a dedicated kit with a validated range in your matrix is usually more accurate, and cross-reactivity in a panel is a real effect.

## Torque calibration

Torque tools drift, and a torque wrench that reads correctly at one setting can be well out at another. A calibration that checks one value in one direction has tested very little, and a certificate without as-found readings has destroyed the evidence you needed. This page covers what to ask for and how to read what comes back.

1. **Ask for several points across the working range**: Tools are not linear across their range and error at the low end of a wrench's scale is common. Points near the bottom, middle and top of the range, rather than a single check at the setting you happen to use, is what characterises the tool.
2. **Require both directions where the tool is used both ways**: A tool used clockwise and anticlockwise needs calibrating in both, because the mechanism is frequently not symmetrical. A certificate covering one direction describes half the tool.
3. **Insist on as-found and as-left values**: The as-found reading is what tells you whether work done since the last calibration was within tolerance. A provider that adjusts before recording has removed the only evidence that answers that question, and that is the single most important line on the certificate.
4. **Check the accreditation scope covers your range**: Accreditation is granted for specific measurements over specific ranges. A provider accredited for torque up to a modest value is not accredited for your large wrench, and the scope document is where that is confirmed.
5. **Set the interval from the tool's own drift history**: Annual is a convention. A tool used constantly, dropped, or historically out of tolerance belongs on a shorter interval, and a tool that never moves can be justified onto a longer one with the evidence to support it.

**What should a torque calibration certificate state?**

The points tested across the range, the directions, as-found and as-left readings, the reference standards and their traceability, the environmental conditions, the measurement uncertainty and the acceptance criteria applied.

**Why do as-found readings matter so much?**

Because they tell you whether work done since the last calibration was within tolerance. If a provider adjusts the tool before recording, that evidence is gone and every assembly since the last visit is unverifiable.

**How often does a torque wrench calibration service say tools need calibrating?**

At an interval justified by the tool's use and its own drift history, reviewed rather than assumed. Heavy use, a drop, or a previous out of tolerance result are all reasons to shorten it regardless of convention.

**Do torque screwdrivers need the same treatment?**

Yes, and low range tools are frequently the ones drifting furthest in relative terms. The same requirements apply: points across the range, both directions where used, as-found values and a stated uncertainty.

## Clinical manufacturing

The step from research material to investigational product is a step into a quality system, and most of the timeline is in things that are not manufacturing: technology transfer, analytical method transfer, release testing, labeling and the slot in somebody else's schedule. This page covers planning it and choosing who does it.

1. **Transfer the process and the analytics together**: A process transfers with its methods, and the methods have to be qualified at the receiving site. Laboratories routinely plan the process transfer and discover the analytical transfer afterwards, which is where several months go.
2. **Budget an engineering batch that produces nothing saleable**: Hold times, filter compatibility, mixing at the new vessel geometry and line speed all have to be re-established at scale. Programs that plan straight into a clinical batch produce an investigation instead of material.
3. **Choose the partner on slots and track record, not on quotation**: Capacity in eighteen months is not capacity, and the format a site runs occasionally is the one that will cause problems. Ask what they run weekly, ask for a reference program of your size, and confirm the slot with penalty terms that mean something.
4. **Plan release, labeling and distribution into the timeline**: Release testing, labeling in the languages and formats the protocol requires, and distribution to sites sit after the batch and routinely add weeks nobody scheduled. Map them before committing to a first patient date.
5. **Treat any process change as a comparability exercise**: A change of site, scale or a raw material has to be shown not to change the product, using the analytical package and, where the change is large, additional characterisation. Comparability planned in advance is a study; discovered late it is a delay.

**What changes when material becomes investigational product?**

It is manufactured under a quality system with a defined process, released against a specification by a responsible person, labeled according to the protocol and tracked to the site. The chemistry may be identical; everything around it is not.

**How do I choose a cdmo partner?**

On what they run routinely at your scale and format, their inspection and quality record, their real slot availability, and the people who would be assigned. Price separates quotations far less than those do.

**How long does technology transfer take?**

Longer than the manufacturing. Process and analytical transfer, an engineering batch and the documentation around them usually dominate the schedule, and the analytical half is the one most often forgotten at planning.

**What is comparability and when do I need it?**

Evidence that a product made after a change is the same as before, from the analytical package and sometimes extra characterisation. You need it whenever site, scale, process or a critical raw material changes, and it is far cheaper planned than discovered.

## Internalisation assays

Internalisation is the property that decides whether an antibody can deliver a payload, and measuring it means separating what went inside a cell from what is still stuck to the outside. Every format is a different way of doing that separation, and the control that makes any of them interpretable is the same one. This page covers the formats and the controls.

1. **Quench the surface signal, the simplest approach**: Label the antibody with a fluorophore, allow internalisation, then add a quenching agent that cannot cross the membrane. Surface signal disappears and internal signal remains. It is straightforward and depends entirely on the quencher being genuinely membrane impermeant under your conditions.
2. **Use a pH sensitive dye, which reports the compartment**: A dye that fluoresces only at acidic pH lights up when the antibody reaches an endosome or lysosome and stays dark on the surface. It needs no quenching step and no washing, which makes it well suited to live cell and kinetic work.
3. **Use a toxin conjugate when function is the question**: A secondary antibody carrying a toxin that only acts inside the cell turns internalisation into cell killing. The readout is functional rather than fluorescent, and it answers the question a payload program actually asks.
4. **Run the four degree binding control on every plate**: At four degrees, antibody binds and does not internalise. That plate is the surface-only reference against which the thirty seven degree plate is read, and without it there is no way to separate internalisation from differences in binding.
5. **Include an isotype control and a non-internalising target**: An isotype control at the same concentration shows the format's background, and an antibody against a target known not to internalise shows the assay can report a negative. An assay that has never produced a negative is not a measurement.

**How do I separate internalised from surface bound antibody?**

By quenching surface fluorescence with a membrane impermeant agent, by using a dye that fluoresces only in acidic compartments, or by an acid wash that strips surface antibody. The four degree control is what anchors whichever method you use.

**Why run the assay at four degrees as well?**

Because antibody binds at four degrees and does not internalise. That gives the surface-only reference, so the difference from the thirty seven degree plate is internalisation rather than a difference in how much bound.

**Flow cytometry or imaging for an antibody internalization assay?**

Flow cytometry for population statistics and speed. Imaging when you need to see where the antibody went, which compartment it reached and whether it recycled. Imaging cytometry gives both at lower throughput.

**Does internalisation rate predict payload delivery?**

It is necessary and not sufficient. Trafficking to the right compartment, payload release and the target's recycling all matter, which is why a functional toxin conjugate assay is the readout that programs eventually rely on.

## Plasmid maxiprep

A maxiprep is a routine procedure that fails in two predictable ways: a low copy plasmid grown in the volume a high copy one needs, and a preparation carrying endotoxin into cells that are then blamed for dying. Neither is visible on a concentration reading. This page covers both and the checks that catch them.

1. **Size the culture from the plasmid's copy number, not the kit's label**: High copy origins give plenty of DNA from a modest culture; low copy and large constructs need several times the volume for the same yield. A kit's stated capacity assumes a high copy plasmid, and using it for a low copy one gives a disappointing yield nobody explains.
2. **Grow it properly and do not over-grow it**: An inoculum from a single colony, selection maintained throughout, and harvest in late exponential phase. Cultures grown far into stationary phase give more biomass, more genomic DNA contamination and often less usable plasmid.
3. **Choose endotoxin free purification when cells or animals are the destination**: Endotoxin from the bacterial outer membrane co-purifies with plasmid and is toxic to primary cells, immune cells and animals. It does not matter at all for a bacterial transformation, which is why the grade is a destination decision rather than a quality one.
4. **Check the supercoiled fraction, not just the concentration**: Run an aliquot on a gel. Supercoiled, nicked and linear forms migrate differently and transfect differently, and a preparation that is largely nicked will transfect poorly at a concentration that reads perfectly well.
5. **Verify the construct on every preparation**: A restriction digest at minimum, and sequencing of the insert and the junctions for anything important. Plasmids recombine and are mixed up, and the cost of verifying is trivial against the cost of a month of experiments on the wrong construct.

**Why is my plasmid maxiprep kit yield low?**

Most often a low copy number plasmid grown in a volume sized for a high copy one, or a culture grown too far into stationary phase. Check the origin of replication on the plasmid map before blaming the kit.

**When do I need endotoxin free plasmid DNA purification?**

Whenever the DNA goes into eukaryotic cells that respond to endotoxin, particularly primary cells and immune cells, or into an animal. For transforming bacteria it makes no difference at all.

**Why does my preparation transfect badly despite a good concentration?**

Often the topology: a largely nicked or linear preparation transfects far worse than a supercoiled one at the same concentration. Run a gel, which takes twenty minutes and answers it.

**Should I verify the plasmid every time?**

A restriction digest on every preparation, and sequencing for anything that will drive a conclusion. Plasmid mix-ups and recombination are common enough that the check pays for itself repeatedly.

## Isocratic and reversed phase

Gradient elution has become the default and it is not always the right answer. Where the analytes span a narrow polarity range, a fixed mobile phase composition gives a shorter cycle, a method that transfers between instruments without a dwell volume argument, and a detector baseline that does not drift. This page covers when to choose it and how reversed phase chemistry underpins both.

1. **Look at the polarity spread of your analytes**: A narrow spread separates well at one composition. A wide spread gives early compounds crowded at the front and late ones broad and retained for an age, which is exactly the problem a gradient exists to solve.
2. **Count the whole cycle, where isocratic wins**: An isocratic method needs no re-equilibration, so the cycle is the run. A gradient method's re-equilibration is frequently a third of the cycle and it is invisible in the chromatogram, which is why gradient methods are slower than they look.
3. **Prefer isocratic where the method must transfer**: Gradient methods carry the instrument's dwell volume into their retention times, so transferring one between instruments requires the dwell volume to be known and matched. An isocratic method has no such dependency, which makes it far more portable.
4. **Start from reversed phase unless there is a reason not to**: A non-polar bonded phase with an aqueous organic mobile phase retains by hydrophobicity and covers the large majority of small molecule work. Normal phase and other modes exist for specific problems, and reversed phase is where a method development starts.
5. **Tune selectivity with the organic modifier and the pH**: Switching between the common organic modifiers changes selectivity rather than just strength, and for ionisable compounds the mobile phase pH relative to the analyte's ionisation is the largest single lever available. Both are worth trying before changing column chemistry.

**When should I use isocratic hplc?**

When the analytes span a narrow polarity range. You get a shorter cycle with no re-equilibration, a flatter baseline and a method that transfers between instruments without matching dwell volumes.

**Why is reversed phase hplc the default?**

Because a non-polar bonded phase with an aqueous organic mobile phase retains by hydrophobicity, which suits the large majority of small molecules, and because the mobile phases are cheap, safe and compatible with mass spectrometric detection.

**How do I change reverse HPLC selectivity without changing column?**

Switch the organic modifier, which alters selectivity rather than only strength, and adjust the mobile phase pH for ionisable compounds. Both are faster and cheaper than buying a different chemistry and frequently sufficient.

**Can I convert a gradient method to isocratic?**

Sometimes. Look at where the peaks elute on the gradient: if they all come off within a narrow composition window, that composition is a good isocratic starting point. If they span the gradient, they will not separate at one composition.

## Cosmetic stability

Cosmetic stability asks a different question from pharmaceutical stability: not only does the active survive, but does the product still look, smell, pour and feel as it should, and does the preservative system still work. It is largely a physical and sensory exercise, and the packaging is part of what is being tested. This page covers the program and the scheduling behind it.

1. **Run accelerated, ambient and cold in parallel**: Elevated temperature accelerates chemical and physical change, ambient is the reference, and cold exposes formulations that crystallise or separate on chilling. A program with only accelerated data misses the cold failures entirely.
2. **Add freeze-thaw cycling, which finds emulsion failures**: Repeated freezing and thawing breaks marginal emulsions and is the fastest way to find a formulation that will not survive a winter delivery van. It is cheap, quick and skipped surprisingly often.
3. **Test light exposure separately**: Color fade and fragrance change under light are separate failures with their own test conditions and their own protected control. For a clear pack on a shelf this is not optional.
4. **Test in the final packaging, both ways up**: The container, the closure and the liner interact with the formulation, and a stability study in a laboratory jar says nothing about a pump pack. Orientation matters where the product contacts the closure, so store upright and inverted.
5. **Run preservative efficacy alongside, not instead**: A preservative challenge test shows the system controls microbial growth, and it should be repeated at the end of the stability period, because preservatives deplete over time and into the packaging.

**What conditions does cosmetic stability testing use?**

Accelerated elevated temperature, ambient as the reference, a cold condition, freeze-thaw cycling and a separate light exposure study with a protected control. The exact temperatures follow the product and the markets it will sit in.

**What is actually assessed?**

Appearance, color, odour, viscosity, pH, phase separation and packaging compatibility, plus preservative efficacy and any actives that need quantifying. Most of it is physical and sensory rather than analytical.

**Does the packaging really have to be the final one?**

Yes. The container, closure and liner interact with the formulation and can absorb fragrance, leach, or let it dry out. A study in a laboratory jar tells you about the formulation and not about the product.

**What does stability testing software need to do?**

Schedule pull points years ahead across many products and conditions, record which chamber each sample is in, capture sensory and instrumental results against the protocol, and warn before a pull is missed. A missed point cannot be recovered.

## Droplet digital PCR

Digital polymerase chain reaction divides a reaction into thousands of tiny partitions, amplifies each one and counts how many were positive. That converts a comparative measurement against a standard curve into a count, which is why it is used for rare variants, copy number and any sample where an inhibitor would distort a quantitative cycle threshold. This page covers what it buys and what it costs.

1. **Partitioning turns concentration into a count**: The reaction is split into many thousands of droplets or wells, each amplifying independently. Counting positives and applying Poisson statistics gives an absolute concentration with no standard curve and no reference material, which is the whole point.
2. **Rare targets are where it wins clearly**: A variant present at a small fraction of a background is hard to distinguish by cycle threshold and straightforward to count when it occupies its own partitions. Rare mutation detection and copy number differences are the established applications.
3. **It tolerates inhibition better**: Because the readout is endpoint presence or absence rather than amplification efficiency, partial inhibition that would shift a cycle threshold has less effect on a count. That matters for difficult matrices where quantitative PCR is unreliable.
4. **The dynamic range is set by the partition count**: Too few target molecules and precision is poor; too many and every partition is positive and the count saturates. The usable range follows from how many partitions the instrument makes, so samples must be diluted into it rather than run as they arrive.
5. **Thresholding is a decision, and it should be a rule**: Separating positive from negative partitions on a fluorescence amplitude plot is where the result is actually made. Set the rule in advance, include a no-template control on every plate, and record the threshold with the data.

**What does a droplet pcr machine do differently from real time pcr machines?**

It partitions the reaction into thousands of independent amplifications and counts the positives, giving an absolute concentration by Poisson statistics rather than a value read off a standard curve.

**When is a digital pcr instrument worth it over a qrt pcr machine?**

For rare variant detection against a background, for small copy number differences, for absolute quantification without a reference standard, and for samples where inhibition makes quantitative PCR unreliable.

**What limits the dynamic range?**

The number of partitions. Below a few target molecules the counting statistics are poor, and above the point where nearly every partition is positive the count saturates. Samples have to be diluted into the usable window.

**Is it more accurate than an rt pcr machine?**

For absolute quantification and small differences, generally yes, because it removes the standard curve and the amplification efficiency assumption. It is also slower, more expensive per sample and has a narrower dynamic range.

## Types of microscopes

Microscope choice is usually presented as a list of instruments and is better read as a list of questions. What is the sample, is it stained, is it alive, how thick is it and how small is the thing you need to see. Answering those five picks the technique, and the instrument follows. This page works through the main types on that basis.

1. **Brightfield, for stained sections and anything with absorption contrast**: The simplest arrangement and the right one for histology, blood films and stained preparations. It is nearly useless for unstained living cells, which have almost no absorption contrast, and that limitation is what every other technique below exists to address.
2. **Phase contrast, for living cells in a culture vessel**: Converts the phase shift a transparent cell causes into visible intensity, working through plastic vessels. It is the default for tissue culture observation and it needs a matched condenser annulus and phase objective rather than being a setting.
3. **Fluorescence and confocal, for specific molecules**: Fluorescence localises labeled molecules; a confocal adds a pinhole that rejects out of focus light, which is what makes optical sectioning of a thick specimen possible. If the specimen is thin and flat, widefield is faster, cheaper and often sufficient.
4. **Stereo, for anything you need to manipulate**: Two optical paths give a three dimensional view at low magnification with a long working distance, which is what dissection, sorting and assembly under the instrument actually require. It is a different job from a compound microscope, not a smaller version of one.
5. **Electron and scanning probe, below the diffraction limit**: Light microscopy stops at around two hundred nanometres. Electron microscopy goes far below that in vacuum on prepared specimens, and atomic force microscopy gives true height data in air or liquid at lower lateral resolution. Both are specialist instruments best accessed through a facility.

**Which of the types of microscopes do I need for cell culture?**

An inverted phase contrast instrument. Cells grow on the vessel bottom with medium above, so the optics must look up through the plastic, and unstained cells need phase contrast to be visible at all.

**Do I need confocal microscopes, or will widefield fluorescence do?**

Widefield for thin, flat specimens such as cultured cells on a coverslip, where it is faster and gentler. Confocal when the specimen is thick and out of focus light is drowning the plane you care about.

**What is the practical resolution limit of a light microscope?**

Around two hundred nanometres for conventional techniques, set by diffraction. Super-resolution methods go below it with specific labeling and acquisition requirements, and electron microscopy goes far below it on prepared specimens.

**Should a laboratory buy phase contrast microscopes or use a facility?**

Buy the routine instrument people use daily, which is usually a phase contrast inverted stand. Use a facility for confocal, super-resolution and electron microscopy, where the instrument is expensive, intermittently used and much better with an expert beside it.

## Tensile testing

A tensile result is a property of a specimen, a standard and a set of grips, and the machine is the least interesting part of that list. Laboratories buy frame capacity and then measure modulus from crosshead travel through a compliant load train, which is the commonest way a perfectly good machine produces numbers nobody else can reproduce. This page covers what to specify.

1. **Size the load cell to the specimens, not to the frame**: Accuracy is a percentage of reading over a working portion of the cell's range, so a small specimen broken on a large cell is measured badly. A frame with interchangeable load cells is usually a better purchase than one big cell.
2. **Choose grips for the material, because slip looks like extension**: Wedge grips for metals, pneumatic grips for films and elastomers, threaded or shouldered fixtures where a specimen must not be crushed. A specimen slipping in the grips produces a curve that looks like yielding and is not, and this is the single commonest error in tensile data.
3. **Measure strain on the specimen with an extensometer**: Crosshead travel includes the compliance of the frame, the grips and the load train, so a modulus calculated from it is wrong and consistently too low. A contact or optical extensometer measuring the gauge length itself is what makes modulus meaningful.
4. **Name the standard, and follow its specimen geometry**: The standard fixes specimen shape, gauge length, preparation, conditioning, strain rate and how properties are calculated. Two laboratories testing the same material to different standards will disagree, and neither is wrong.
5. **Condition the specimens and record the environment**: Polymers in particular are sensitive to temperature and humidity, and standards specify a conditioning period for that reason. A laboratory that does not record its test environment cannot explain a seasonal drift in its own results.

**How do I size a tensile testing machine?**

From the highest force your specimens require, with a frame that accepts interchangeable load cells so small specimens are measured on a small cell. Accuracy is a percentage of reading, so results should sit in the middle of the cell's range.

**Why is my modulus lower than the published value?**

Almost always strain measured from crosshead travel rather than from the specimen. That includes the compliance of the frame and grips, which inflates apparent extension and depresses modulus. Use an extensometer.

**What causes slipping in the grips?**

Wrong grip type for the material, insufficient clamping pressure, or a specimen surface that is too smooth. Slip shows as a curve with a false yield and it invalidates the result, which is why specimens should be examined after a break.

**Does the strain rate matter?**

Substantially for polymers and for anything rate sensitive, which is why standards specify it. A faster test gives a higher apparent strength, so comparing results run at different rates compares the rates.

## Vibration and battery testing

Vibration testing reproduces a service environment on a shaker table, and the two things that decide whether the result means anything are the profile and the fixture. Battery work adds a third: the samples are energetic, and abuse testing is done behind containment by people equipped for a thermal event. This page covers commissioning both.

1. **Name the profile, the axes and the duration**: A random vibration profile, a sine sweep or a shock pulse are different tests, and the level, the frequency range, the duration and which axes are driven all have to be stated. A request to vibration test something is not a specification.
2. **Treat the fixture as part of the test**: The fixture couples the shaker to the specimen and its own resonances will either amplify or absorb the input. A fixture designed to be stiff through the test frequency range, with control accelerometers placed as the standard requires, is what makes the input the specimen sees the input you asked for.
3. **Decide what constitutes a failure before the test**: Visible damage, loss of function, an intermittent connection during the test, or a change in a resonance survey before and after. Deciding afterwards is deciding with a result in front of you, and a resonance survey either side is the most informative comparison available.
4. **For batteries, establish the containment and the safety case first**: Abuse testing deliberately drives cells to failure, so the laboratory needs a containment chamber, gas handling, fire suppression and people trained for a thermal runaway. Ask to see the facility rather than reading about it.
5. **Check the accreditation scope covers the exact test**: Accreditation is granted for specific standards over specific ranges and axes, and transport testing in particular is governed by specific regulations. A laboratory accredited for one standard is not thereby accredited for the one your product needs.

**What has to be specified for vibration testing?**

The standard or profile, the level and frequency range, the duration, which axes are driven, the mounting and the pass criteria. Without those, two laboratories will run different tests and both will be correct.

**Why does the fixture matter so much?**

Because it sits between the shaker and the specimen and has its own resonances. A poor fixture amplifies some frequencies and absorbs others, so the specimen never sees the specified input and the test proves nothing.

**What does a battery testing laboratory need that others do not?**

Containment chambers, gas extraction, fire suppression and staff trained for thermal runaway, because abuse testing deliberately causes failure. For transport testing, accreditation to the specific regulations that govern shipping cells.

**Should the product be tested functioning?**

Where it is functioning in service, yes, with monitoring during the test. Intermittent faults that appear only under vibration are exactly what the test exists to find and are invisible in a post-test inspection.

## Polarised light

Polarised light microscopy reports on molecular order rather than on chemistry. A crystalline or aligned material rotates polarised light and appears bright against a dark field between crossed polarisers, and a compensator turns that brightness into a signed measurement. It is a classical technique, it is cheap to add to a good stand, and it is unforgiving of strained optics and thick sections.

1. **Cross the polariser and the analyser, and check the extinction**: With no sample, a properly crossed pair gives a dark field. Poor extinction means the pair is not crossed or the optics are strained, and a system that cannot reach a good extinction cannot reliably report weak birefringence.
2. **Use strain-free optics**: Ordinary objectives carry internal stress that itself rotates polarised light, producing a background that mimics the sample. Objectives and condensers designated strain-free exist for this reason and are not an optional refinement for quantitative work.
3. **Add a compensator to turn brightness into a signed result**: A first order or a variable compensator shifts the retardation so that a birefringent object appears in a color or intensity that reports the sign of its elongation. That sign is the actual diagnostic information in several classical applications.
4. **Rotate the stage, because orientation is the measurement**: A birefringent object goes dark four times in a full rotation, at the positions where its axes line up with the polarisers. Reporting a birefringent object without rotating the stage risks calling an aligned object isotropic.
5. **Control section thickness, because retardation scales with it**: Retardation is the product of birefringence and path length, so a thicker section gives a stronger signal from the same material. Comparing objects across sections of different thickness compares the sectioning.

**What does polarized light microscopy show?**

Birefringence, which is a property of ordered material: crystals, aligned fibres and some biological structures rotate polarised light and appear bright between crossed polarisers, where isotropic material stays dark.

**Why do I need strain-free objectives?**

Because ordinary lenses carry internal stress that rotates polarised light themselves, producing a background indistinguishable from a weakly birefringent sample. For any quantitative work, strain-free optics are a requirement.

**What is a compensator for?**

It adds a known retardation so that the sign and magnitude of the sample's birefringence can be read rather than only its presence. That signed result is the diagnostic information in several classical identifications.

**Why does my sample disappear when I rotate the stage?**

Because a birefringent object extinguishes when its optical axes align with the polariser and analyser, four times in a full rotation. That is the expected behavior and it is how orientation is measured.

## High throughput cytometry

Running cytometry from plates changes the bottleneck. Acquisition is rarely the limit; sample settling in the wells, carryover between them, and the volume of data and analysis are. A screen that ignores the first two produces a plate effect that looks like a hit. This page covers building a throughput workflow that stays interpretable.

1. **Specify the loader on mixing and wash, not on speed**: Cells settle within minutes, so a loader that mixes immediately before each acquisition is the difference between a real gradient across the plate and a measured one. A backflush or wash between wells is what controls carryover.
2. **Measure carryover before trusting a screen**: Run a bright sample followed by a blank, repeatedly, and quantify what appears in the blank. On a screen with a wide dynamic range, a small carryover fraction from a strong well is indistinguishable from a weak positive in the next.
3. **Randomise the plate layout**: Controls and treatments in fixed columns turn any edge, settling or carryover effect into a systematic difference between groups. Randomising the layout and recording it is what allows a plate effect to be detected and modelled.
4. **Put controls on every plate, not on a control plate**: An unstained well, a single stained set for the panel and a known positive and negative on each plate. A screen whose compensation controls were run once on plate one carries that day's settings across everything.
5. **Plan the analysis and the storage before the first plate**: Thousands of files need automated gating with a defined template, quality control that flags wells with too few events or a clog, and storage for the raw files. Gating a screen by hand is how a two-week screen becomes a two-month one.

**What actually limits high throughput flow cytometry?**

Sample settling in the plate, carryover between wells, and the analysis. Acquisition speed is rarely the constraint, and a loader chosen on speed alone will produce a plate gradient that looks like biology.

**How do I measure carryover?**

Alternate a bright sample with a blank across a plate and quantify what appears in the blanks as a fraction of the bright well. Do it at the dynamic range your screen will actually span, because carryover matters most when wells differ greatly.

**Do I need controls on every plate?**

Yes. An unstained well, single stained controls for the panel, and known positive and negative samples. Compensation and thresholds drift across a screen, and per-plate controls are what let that be corrected rather than ignored.

**How should the data be gated?**

With a defined template applied automatically, plus a quality control pass that flags wells with low event counts, clogs or abnormal scatter. Manual gating of thousands of wells is both slow and a source of drift.

## ELISpot readers

An ELISpot result is a number produced by an algorithm from an image, and the parameters that algorithm uses decide the number as much as the biology does. Two readers, or one reader with two parameter sets, will report different counts from the same plate. This page covers choosing a reader and locking the analysis so a study's counts mean one thing throughout.

1. **Judge it on the counting, not the optics**: Every reader images a well adequately. What separates them is how the software separates a genuine spot from a speckle, how it handles merged spots in a high responder, and how reproducibly it does both. Ask to run your own plates, including a high responder.
2. **Lock the counting parameters before the study starts**: Size, intensity and gradient thresholds decide the count. Set them on representative plates, freeze them, and apply the same template to every plate in the study. Adjusting parameters per plate is adjusting the result.
3. **Require a per-well image that a person can audit**: The count has to be checkable. A reader that stores the image with the count lets somebody look at a surprising well; one that reports only numbers leaves an outlier unexplainable.
4. **Put a positive control on every plate**: A polyclonal stimulus well on each plate shows the cells responded and the detection worked. A plate with no response and no positive control cannot be distinguished from a plate where the assay failed.
5. **Standardize plate handling, because it shows up in the count**: Washing, drying and the time between development and reading all change spot appearance. Dry plates fully and read them in a consistent window, and record the reading date with the count.

**What actually separates one elispot plate reader from another?**

The counting software. Spot detection, how merged spots in a strong responder are resolved and how reproducible the count is between reads are what differ; the imaging is adequate on all of them.

**Why do two readers give different counts?**

Because different algorithms and parameter sets separate spots from background differently. Counts from two systems are not interchangeable, which is why a study should use one reader and one locked parameter template.

**Should counting parameters be adjusted per plate?**

No. Set them on representative plates before the study and freeze them. Adjusting per plate means the analysis responds to the result, which is the hardest kind of bias to detect afterwards.

**What controls does an ELISpot plate need?**

A negative control well with no stimulus, a positive polyclonal stimulus well, and, for a study, the same donor control sample across plates. Without the positive control a negative result is uninterpretable.

**How much of a plate reader assay is the instrument?**

Less than people expect. The plate, the incubation, the blocking and the operator's timing move a result more than the optics do, which is why two laboratories with identical readers disagree. Standardize the assay first and treat the reader as the last thing to change.

**How does a microplate reader vs spectrophotometer comparison come out?**

They measure the same physics through different optics. A cuvette instrument has a long fixed path and gives the better absolute absorbance on one sample; a plate reader has a short path set by the liquid volume and gives you ninety six comparable relative readings. So a concentration from first principles belongs on the cuvette instrument, and anything with a standard curve across a plate belongs on the reader.

## Pipette controllers

A pipette controller is the instrument nobody specifies and everybody uses, and the differences that matter are not in the motor. Speed control decides whether a cell suspension can be layered gently, the filter decides whether the instrument survives an aspiration accident, and battery life decides whether it works at the end of the afternoon. This page covers all three.

1. **Insist on proportional speed control in both directions**: Layering a suspension over a gradient, or dispensing the last millilitre of a culture without shearing it, needs fine control at low speed. A controller with two speeds is adequate for filling flasks and frustrating for anything gentle.
2. **Check the filter and how easily it is changed**: A hydrophobic filter, usually rated at nought point two two micrometres, keeps aerosol and liquid out of the instrument and keeps the instrument out of a sterile workflow. Filters block and get wet, so the change should be a ten second job without tools.
3. **Match the pipette range and the sealing nozzle**: Controllers handle a stated range of serological pipette sizes, and the nozzle has to seal on all of them. A small pipette in an instrument built for large ones leaks, which presents as poor volume control rather than as a fit problem.
4. **Judge the battery on a working day, not on a specification**: Runtime, whether it charges on a stand and whether it can be used while charging decide whether it works in the afternoon. A controller that has to be taken out of the cabinet to charge is a controller somebody will be waiting for.
5. **Handle it for weight and balance before buying**: This instrument is held for hours. Weight, balance and where the thumb sits are the specification that people actually feel, and they cannot be read off a datasheet. Borrow demonstration units and let the users decide.

**What does a serological pipette aid do?**

It provides the suction and pressure for a serological pipette under motorised control, replacing a rubber bulb. The gain is fine speed control in both directions, which is what makes gentle layering and controlled dispensing possible.

**Why does the filter matter?**

It stops aerosol and accidentally aspirated liquid entering the instrument, protecting both the mechanism and the sterility of the work. Filters block and wet out, so a design that changes them in seconds matters more than the filter itself.

**Is an auto pipette worth it over a bulb?**

For any sustained serological work, clearly. A bulb is slow, tiring and gives poor control, and repetitive strain from bulb use is a real occupational problem rather than a theoretical one.

**What goes wrong with pipette controllers?**

A blocked or wet filter, a worn nozzle seal and a tired battery, in that order. All three are consumables, all three present as poor suction, and all three are commonly diagnosed as a failed instrument.

**What matters most in a pipetting controller?**

Proportional speed control in both directions, a filter that is easy to change, a seal that matches the sizes you use, and a battery that lasts a working day. The motor itself is rarely the differentiator.

## Cell cycle markers

Two of the most used antibodies in cell biology report things people routinely over-read. A proliferation marker stains cells that were recently cycling as well as those cycling now, and a cytoskeletal marker is exquisitely sensitive to how the cell was fixed. Both are excellent reagents used carelessly. This page covers what each reports and the preparation each needs.

1. **A proliferation marker reports recent history, not a snapshot**: The protein accumulates through the cell cycle and persists after the cell leaves it, so staining reports cells that have cycled recently as well as those cycling now. Where the question is which cells are in S phase at this moment, a nucleotide incorporation assay answers it and an accumulation marker does not.
2. **Choose the proliferation marker to match the question**: Different proliferation antigens have different persistence and different cell cycle profiles, so they give different indices on the same tissue. Report which one was used and do not compare an index measured with one against a published figure measured with another.
3. **Handle cytoskeletal staining as a fixation problem**: Microtubules depolymerise in the cold and are poorly preserved by some fixatives. Aldehyde fixation at temperature, or a methanol protocol, each preserve different aspects, and a disappointing cytoskeletal image is nearly always fixation rather than antibody.
4. **Anchor every count to a nuclear stain**: A proliferation index is a fraction, so the denominator has to be counted. A general nuclear marker counterstain in a separate channel is what makes the fraction measurable rather than estimated, and it also exposes tissue folds and debris.
5. **Run the controls that make the index credible**: A tissue known to be proliferating, one known not to be, and a no-primary control on the same run at the same settings. An index reported without them is an impression with a number attached.

**What does a pcna antibody actually show?**

A protein that accumulates during the cell cycle and persists for a period afterwards, so it marks recently and currently cycling cells. It is an accumulation marker, not a snapshot of cells in S phase.

**Are proliferation markers interchangeable?**

No. They differ in cell cycle profile and in how long they persist, so the index they give on the same tissue differs. Always state which marker produced an index, and do not compare across markers.

**Why does my tubulin antibody give a poor image?**

Fixation, almost always. Microtubules depolymerise in the cold and some fixatives preserve them badly. Test an aldehyde protocol at temperature against a methanol protocol on a known positive before changing antibody.

**Do I need a nuclear marker antibody or counterstain?**

For anything quantitative, yes. A proliferation index is a fraction and the denominator has to be counted, which needs a general nuclear marker in its own channel. It also reveals folds and debris that would otherwise be counted.

## Tumour target antibodies

These three targets are grouped here because they are bought for the same reason: each is a cell surface antigen under active therapeutic interest, so the catalog is full of reagents whose descriptions borrow clinical language. What matters when buying one is more ordinary than that, and it is the same for all three: which epitope, which application, and what you will compare the stain against.

1. **Match the epitope to the application**: An antibody against the extracellular domain detects the receptor on intact cells by flow cytometry; one against an intracellular region does not, and will only work on permeabilised or lysed material. This is the most common mismatch on surface targets and it is visible on the datasheet.
2. **Ask whether it blocks, if blocking is the point**: Detection and function blocking are different products. A reagent claimed to block should show a dose response in a functional assay; binding data alone does not demonstrate it, and the two are frequently presented side by side.
3. **Use expression-defined controls, not tissue impressions**: A cell line known to express the target strongly and one known not to, run alongside every stain, is what turns an image into evidence. For a surface target these lines are usually available and the control costs one well.
4. **Quantify receptor density where you will compare lines**: The same antibody gives very different apparent staining on cells expressing different amounts of the target, so comparisons between models need a density measurement rather than an impression. Calibrated beads make this straightforward on a cytometer.
5. **Specify the grade for what it will touch**: Carrier-free, low endotoxin and azide-free for anything added to living cells; carrier-free also for anything you intend to conjugate. A general purpose preparation with carrier protein cannot be labeled and should not be added to cells.

**Why does my B7H3 antibody not stain the cell surface?**

Most often the epitope is intracellular. Check which region the immunogen came from: an extracellular domain antibody detects intact cells, an intracellular one requires permeabilisation and will look like a failed stain otherwise.

**How do I know an antibody blocks rather than just binds?**

A functional assay with a dose response and a stated potency in a named system. Binding affinity, however good, is not evidence of blocking, and many catalog entries present the two together.

**What controls does a tumour surface target stain need?**

A line known to express the target and one known not to, stained in the same run at the same settings, plus an isotype control. Those three separate a real result from an impression.

**Can I compare staining intensity between cell lines?**

Only with a receptor density measurement, usually with calibrated beads on a cytometer. Apparent intensity depends on expression level, and comparing lines without quantifying it is how internalisation and efficacy results fail to replicate.

## Neuro target antibodies

These three targets break the default protocol in ways that present as a failed antibody. One is a multipass membrane channel that aggregates when boiled, one is a cytoplasmic ligase whose detection is complicated by its own modification, and one is an enormous channel that will not enter an ordinary gel. Knowing which problem you have before buying saves weeks.

1. **Do not boil a multipass membrane protein**: Heating a multipass membrane channel in sample buffer aggregates it, and the aggregate sits in the well rather than running. Incubating at a lower temperature, commonly thirty seven or seventy degrees, is what makes the band appear, and this single change rescues a great many apparently dead antibodies.
2. **Use a low percentage gel for very large targets**: A protein of several hundred kilodaltons will not resolve on a standard gel and may barely leave the stacking layer. A low percentage or gradient gel, a longer run and a wet transfer with extended time are what get it onto a membrane.
3. **Check whether the epitope is conformational**: Antibodies raised against folded extracellular loops often fail on a denatured blot and work well on intact cells. Read which immunogen was used and which applications carry actual data, because the application list is frequently aspirational.
4. **Expect modification to change the picture**: Ubiquitination and other modifications produce ladders and shifted bands that look like non-specific binding. For targets in that biology, a smear or a ladder may be the real signal and a single clean band may be the artefact.
5. **Use a genetic negative wherever one exists**: A knockout or knockdown sample is the control that settles specificity for all three. Where none exists, an overexpression positive and a second antibody against a different epitope together are the next best evidence.

**Why does my membrane protein not appear on a western blot?**

Most often because the sample was boiled and the protein aggregated in the well. Try thirty seven or seventy degrees instead of ninety five, and check whether the antibody's epitope is conformational and therefore lost on a denatured blot.

**How do I blot a very large protein?**

A low percentage or gradient gel, a long run, and a wet transfer with extended time and often reduced methanol. Standard conditions leave proteins of several hundred kilodaltons in the gel, which reads as an absent target.

**What does a conformational epitope mean in practice?**

The antibody recognizes the folded shape, so it works on intact cells and fails on a denatured blot. Applications are not interchangeable for such a reagent and the datasheet should show data per application.

**Why is my blot a ladder rather than a band?**

For targets in ubiquitin biology, a ladder is often the real signal rather than non-specific binding. Interpreting it needs the genetic control and a treatment that changes the modification, not a cleaner antibody.

## CD marker antibodies

CD numbers name an antigen, not a reagent. Two antibodies against the same CD antigen can recognize different epitopes, behave differently after fixation, and give quite different separation between positive and negative populations. What you are buying is the clone and its conjugate, and this page covers how to choose between them and what each of these six markers is actually reporting.

1. **Buy the clone, not the CD number**: The antigen is defined by the CD nomenclature and the reagent is defined by the clone. Published panels cite clones for that reason, and substituting a different clone against the same antigen is a change that has to be re-titrated and re-validated rather than assumed equivalent.
2. **Know what each marker actually resolves**: CD45 separates leucocytes from everything else and is the usual backbone of a panel. CD19 marks B cells, CD3e marks T cells, CD14 marks monocytes and CD68 marks macrophages, largely intracellularly. CD44 is broadly expressed and is a state marker rather than a lineage one, which is why it needs a quantitative rather than a positive or negative reading.
3. **Match the format to the assay**: A clone selected against native surface protein for flow cytometry may not work on a fixed section, and an antibody validated for immunohistochemistry may not stain live cells. Read the evidence per application rather than the application list, and buy against the assay you will run.
4. **Titrate every conjugate on your own sample**: The datasheet dilution came from somebody else's cells and buffer. Run a dilution series and choose the concentration giving the best separation between positive and negative, which is frequently several times more dilute than recommended and always cheaper.
5. **Assign brightness against abundance across the panel**: Put the brightest fluorochromes on the least abundant antigens. A very bright conjugate on an abundant marker like CD45 spreads into neighbouring detectors and costs resolution on the markers that needed it.

**Are two antibodies against the same CD antigen interchangeable?**

No. They may recognize different epitopes, tolerate fixation differently and separate populations differently. The clone is the product, which is why published panels cite clones rather than CD numbers alone.

**What does a cd45 antibody do in a panel?**

It separates leucocytes from debris, red cells and non-haematopoietic material, which is why it is the usual backbone marker. Getting it right makes every other gate in the panel cleaner.

**Why is a CD44 antibody harder to interpret than a lineage marker?**

Because it is broadly expressed and its level changes with cell state rather than identity. It should be read as a quantitative measure against a control, not as a positive or negative gate.

**Do I need to re-titrate when the lot changes?**

Yes, and it takes an afternoon. Conjugate lot changes shift the optimal concentration, and a panel that drifts across a study is usually a lot change nobody re-titrated.

**Can I use a research clone as a therapeutic comparator?**

As a positive control or a blocking reagent, yes. As an equivalent product, no: formulation, endotoxin specification and release testing differ, and only research use is supported.

**Can one marker identify a population?**

No. Markers are shared across lineages, appear on activation and disappear with maturation. A lineage is a combination including exclusions, written down as a gating sequence with its clones.

**Can I translate a human panel to mouse?**

Not marker by marker. Different markers define the same populations in the two species and some do not exist in the other. Use the established panel for the species and confirm against a known control tissue.

## Neural marker antibodies

Neural tissue is where marker antibodies are most often over-read, because a label that marks a compartment gets reported as marking a cell type and a label that rises with activation gets reported as marking cell number. These four are widely used, each reports something specific, and each has a preparation that makes or breaks it.

1. **MAP2 labels the somatodendritic compartment**: It marks the cell body and dendrites and excludes the axon, which is precisely what makes it useful for distinguishing dendritic from axonal processes in culture. It is not a count of neurons and it is degraded by ischaemia, so post-mortem interval matters in tissue.
2. **Nestin marks progenitors and reactive states, not a lineage**: An intermediate filament of neural progenitors that is re-expressed in reactive astrocytes and in endothelium. Reporting nestin as a progenitor count without excluding vessels and reactive glia is a common overreach.
3. **NeuN is a nuclear neuronal marker with known exceptions**: It labels most neuronal nuclei and is the usual denominator for neuronal counts, and there are well documented neuronal populations it does not label. Where the count matters, state which populations were counted rather than implying all neurons.
4. **SOX10 is a nuclear glial and neural crest marker**: Nuclear staining makes it straightforward to count, which is its practical advantage over cytoplasmic glial markers. It also marks melanocytic lineage, which matters in tumour work.
5. **Test retrieval before changing antibody**: Nuclear markers generally need heat induced retrieval and cytoskeletal ones frequently do not, and over-fixed archival tissue needs harder conditions. Run citrate near pH six against a Tris buffer near pH nine on a known positive section before concluding a reagent does not work.

**Does a map2 antibody label axons?**

No, and that is the point. It marks the cell body and dendrites and is excluded from the axon, which is what makes it the standard way to distinguish dendritic from axonal processes in culture.

**Can I count neurons with a neun antibody?**

For most populations, yes, and there are documented neuronal types it does not label. Report which regions and populations were counted rather than implying a total neuronal count.

**Is a nestin antibody a progenitor marker?**

It marks progenitors and is also re-expressed in reactive astrocytes and in endothelium. A nestin count without excluding vessels and reactive glia overstates the progenitor population.

**Why does my staining work in culture and fail in tissue?**

Fixation and retrieval. Cultured cells are lightly fixed and accessible; tissue is cross-linked and needs retrieval matched to the epitope. Test two retrieval buffers on a known positive before changing the reagent.

**What does a SOX10 antibody label?**

Glial and neural crest nuclei. Because the stain is nuclear it is straightforward to count, which is its practical advantage over cytoplasmic glial markers, and it also marks the melanocytic lineage, which matters in tumour work.

## Cytokine antibodies

Cytokine antibodies are bought for three different jobs that a catalog entry frequently blurs: detecting the protein, capturing it in an assay, and blocking its function in a cell experiment. Each needs different evidence, and several of these targets exist as a precursor and a processed mature form that different antibodies see differently. This page covers choosing for each job.

1. **Decide detection, capture or neutralisation first**: A detection antibody binds on a blot or a section. A capture antibody works as half of a matched pair in an immunoassay. A neutralising antibody blocks a function and must show a dose response in a cell assay. These are three products and a datasheet that lists all three needs evidence for all three.
2. **Establish which form the antibody sees**: Several of these cytokines are made as a precursor and cleaved to a mature form, and some circulate bound to soluble receptors. An antibody against the precursor region reports something different from one against the mature protein, and this explains a great deal of disagreement between published measurements.
3. **Use matched pairs for immunoassay, from one supplier**: Capture and detection antibodies must bind different epitopes without blocking each other. Assembling a pair from two independently chosen antibodies is a development project; buying a validated pair is a purchase.
4. **Specify carrier-free and low endotoxin for cell work**: Anything added to living cells needs a stated endotoxin limit, no azide and no bovine carrier protein. Endotoxin in a cytokine experiment is a particularly bad contaminant, because it induces the very cytokines you are measuring.
5. **Check species reactivity was tested, not predicted**: Cytokines are less conserved than their names suggest, and cross-reactivity predicted from sequence often fails in practice. For a model organism, ask for data in that species and for neutralisation specifically, where potency differs even when binding does not.

**Is a tnf alpha antibody the same as a tnf-alpha antibody?**

The hyphen is a naming convention rather than a difference, and suppliers use both. What differs between products is the clone, the epitope and whether the reagent was validated for detection, capture or neutralisation.

**How do I know an antibody neutralises?**

A cell-based functional assay with a dose response and a stated half-maximal inhibitory concentration in a named system. Binding data, however strong, is not evidence of blocking and the two are frequently shown together.

**Why do two IL-1b antibody assays disagree about a cytokine concentration?**

Different antibody pairs see different forms, including precursor, mature and receptor-bound cytokine, and standards are calibrated differently. Values from two kits are generally not interchangeable.

**What grade do I need to add antibody to cells?**

Carrier-free, azide-free, sterile filtered and with a stated endotoxin limit. Endotoxin is the contaminant that matters most here because it stimulates cytokine production directly.

**Why do two IL-6 antibody immunoassays report different concentrations?**

Usually because they are calibrated against different recombinant standards, and often because they detect different forms of the protein. Compare assays only when the standard and the forms detected are both stated.

## Organelle markers

Organelle markers are used for two quite different purposes and confused between them constantly: as a reference that says which compartment a fraction or an image region is, and as a readout of an organelle's behavior. The first is a control and the second is an experiment, and reading a control as an experiment produces some of the most persistent errors in cell biology. This page separates them.

1. **Use a compartment marker as a reference, not as a quantity**: An endoplasmic reticulum marker tells you a fraction contains that compartment or an image region is that compartment. It does not measure how much endoplasmic reticulum a cell has, and treating a change in marker level as a change in organelle content requires far more evidence than a blot.
2. **Run the full marker panel on a fractionation**: A fraction is characterised by what is enriched and by what is depleted, so markers for the other compartments have to be blotted alongside. A fraction claimed to be one organelle with only that organelle's marker shown is an unverified claim.
3. **Read LC3 as a conversion, not as a level**: The autophagy marker exists as a soluble and a lipidated form that run at different sizes, and the ratio between them is the readout. Both an increase in autophagy and a block in degradation raise the lipidated form, which is why the measurement requires a lysosomal inhibitor arm.
4. **Load equally and normalise to total protein**: For markers used as loading or fractionation references, total protein staining of the membrane is a better normaliser than another single protein, because a single reference can itself change with treatment.
5. **Validate the marker in your own cell type**: Organelle marker abundance varies widely between cell types, and a marker that is strong in a secretory cell may be faint elsewhere. Establish the expected pattern in an untreated sample before interpreting any change.

**What does a calnexin antibody mark?**

The endoplasmic reticulum membrane. It is used as an ER reference in fractionation and imaging, and as a loading control in some contexts, rather than as a measure of how much ER a cell contains.

**How should an lc3a antibody result be read?**

As the ratio between the soluble and lipidated forms, in the presence and absence of a lysosomal degradation block. Without the block, an increase in the lipidated form cannot distinguish more autophagy from less degradation.

**Can an endoplasmic reticulum marker antibody be a loading control?**

Only if it is genuinely invariant under your treatment, which has to be shown rather than assumed. Total protein normalisation avoids the question and is generally the better answer.

**Why does my fractionation look impure?**

Because fractionation is enrichment rather than purification, and adjacent compartments co-sediment. Blot the markers for the other compartments and report enrichment, rather than claiming purity nobody achieved.

**Can I stain lipids with a standard immunofluorescence protocol?**

No. The usual permeabilisation extracts much of the target. Lipid staining protocols specify their own fixation and permeabilisation, and a protein protocol will report an absence that is an artefact of the method.

## Transcription factor antibodies

Transcription factors are low abundance, nuclear and frequently regulated by modification rather than by amount, which makes them the hardest class of antibody target in routine use. A protocol built for a cytoplasmic protein will report that the factor is absent. This page covers what a nuclear target actually needs and what evidence to demand before buying.

1. **Use a permeabilisation that reaches the nucleus**: The detergent and buffer that permeabilise a cytoplasmic protein frequently do not open the nuclear envelope adequately. Nuclear factor protocols use their own buffer systems, and for flow cytometry a dedicated nuclear permeabilisation kit is the practical answer rather than an adjustment.
2. **Expect low abundance and plan the amplification**: Many of these proteins are present at a small fraction of a structural protein's level, so a direct conjugate may simply not reach. Amplification through a biotinylated intermediate or a polymer detection system is what makes the signal visible, at the cost of harder quantitation.
3. **Insist on a genetic negative, because isotype controls are weak here**: An isotype control shows the format's background and says nothing about specificity for a nuclear protein at low abundance. A knockout, a knockdown or a cell type known not to express the factor is what separates signal from nuclear background.
4. **Know whether the epitope is modification dependent**: Several of these targets are read through their modifications, and an antibody against a modified residue will not see the unmodified protein and vice versa. Read the immunogen and ask for the specificity evidence, ideally a peptide array.
5. **Match the reagent to the application, not the target**: An antibody validated for chromatin immunoprecipitation has been shown to work on fixed, fragmented chromatin, which is a different demonstration from a western blot. Buy against the application you will run and read that application's evidence specifically.

**Why can I not detect my transcription factor?**

Usually abundance and permeabilisation. These proteins sit far below structural proteins in level and need a nuclear permeabilisation protocol; a cytoplasmic protocol reports absence rather than a low signal.

**Is an isotype control enough for a nuclear stain?**

No. It shows the background of the reagent format and nothing about specificity at low abundance in a nucleus full of protein. A genetic negative, or a cell type known not to express the factor, is what settles it.

**What does a foxp3 antibody flow cytometry panel need that a surface panel does not?**

A nuclear fixation and permeabilisation system, applied after the surface markers are stained, and surface markers that survive it. Several surface epitopes are damaged by nuclear permeabilisation buffers.

**How do I judge the specificity of epigenetic antibodies?**

By evidence that it discriminates between neighbouring modification states, which means a peptide array or a dot blot against a panel of modified peptides, plus a peptide competition you can run yourself.

**Can one antibody distinguish between family members?**

Only if the immunogen sits in a divergent region and the supplier shows it was tested against the other members. Where that distinction carries the experiment, confirm it in cells lacking each member rather than relying on a single clean band.

**Why does my nuclear target give no band at all?**

Most often because it was never in the lysate. Chromatin-bound factors need an extraction that releases them, so confirm the fractionation with a nuclear marker before concluding the antibody has failed.

## Transporter antibodies

Transporters and multipass receptors are the targets most likely to make a good antibody look dead. They aggregate when boiled, need a detergent that solubilises without destroying the epitope, and are frequently glycosylated enough to run as a smear rather than a band. None of that is the antibody's fault, and all of it is predictable. This page covers the preparation each demands.

1. **Do not boil a multipass membrane protein**: Heating in sample buffer aggregates multipass proteins, and the aggregate stays in the well. Incubating at thirty seven or seventy degrees instead is the single change that rescues most apparently failed transporter blots, and it should be the first thing tried.
2. **Choose the detergent for solubility and for the epitope**: A detergent that extracts the protein may denature the epitope, and a gentle one may leave the target in the pellet. Test two or three on a known positive lysate and look at both the pellet and the supernatant before concluding anything.
3. **Expect glycosylation to broaden the band**: Heavily glycosylated transporters run as a smear well above the predicted mass. Enzymatic deglycosylation collapses it and confirms identity, and it is the fastest way to show that a smear is your protein rather than background.
4. **Use a genetic negative, and an overexpression positive**: For a membrane protein at native abundance, a knockout or knockdown lysate is the control that settles specificity. An overexpression lysate as a positive tells you the antibody can see the protein at all, which is worth knowing before hunting for it at endogenous level.
5. **Check whether the epitope is intracellular or extracellular**: For surface staining of intact cells the epitope must be on an extracellular loop, which for a multipass transporter is a short and awkward region. Antibodies against intracellular loops are common and will not stain an intact cell whatever the protocol.

**Why does my transporter antibody show nothing on a blot?**

Most often the sample was boiled and the protein aggregated in the well, or the detergent left it in the pellet. Try thirty seven or seventy degrees and check the pellet before changing reagent.

**Why is the band a smear?**

Glycosylation. Heavily glycosylated transporters migrate broadly and above the predicted mass; enzymatic deglycosylation collapses the smear to a band and confirms the identification.

**Can I stain a transporter on intact cells?**

Only with an antibody against an extracellular loop. Most catalog reagents for these targets are raised against intracellular regions, which is visible on the datasheet and is the commonest reason a surface stain fails.

**What control settles specificity?**

A knockout or knockdown sample. For membrane targets at native abundance the background is high enough that an isotype control proves very little, and a genetic negative is the only strong evidence.

**Why does my transporter run at an unexpected size?**

Heavy glycosylation shifts apparent mass, and heating can aggregate the protein so that it never enters the gel. Use a gentle solubilisation without boiling, enrich the membrane fraction, and validate against cells that do and do not express it.

## Cell death antibodies

Cell death markers are read wrongly more often than almost any other class, because the informative measurement is nearly always a cleavage or a relocation rather than an amount. Total protein does not change when a cell dies by these routes; the processed fragment appears and the protein moves. This page covers what each marker reports and how to control for a culture that was simply unhealthy.

1. **Read the cleaved form, not the total**: The executioner caspase is present as an inactive precursor in healthy cells and is activated by cleavage. An antibody against the total protein shows no change during apoptosis; one specific for the cleaved neoepitope is the measurement. Buy the cleavage-specific reagent and say which you used.
2. **Read gasdermin the same way**: The pore-forming fragment is generated by cleavage, and the full length protein sits in healthy cells. A blot showing total gasdermin tells you the cell has the protein; the cleaved band tells you the pathway ran.
3. **Read HMGB1 by location, not level**: It is a nuclear protein that relocates to the cytoplasm and is released, so the informative measurements are a nuclear to cytoplasmic ratio by imaging or fractionation, and its appearance in the medium. A whole-cell level barely moves.
4. **Include a positive control that definitely triggers the pathway**: A known inducer at a known concentration, run alongside, is what shows the assay can detect the event. A negative result with no positive control is uninterpretable, and cell death assays are where that omission is most common.
5. **Separate specific death from a sick culture**: Confluence, medium exhaustion, handling and a poor thaw all cause background death that looks like a treatment effect. Include a vehicle control from the same plate and a viability measure, and report the baseline rather than only the difference.

**Should I buy a total or a cleaved caspase 3 antibody?**

Cleaved, for measuring apoptosis. The precursor is present in healthy cells, so a total antibody shows little change; the cleavage-specific reagent recognizes the neoepitope created by activation.

**How is an hmgb1 antibody result read?**

By location. It is nuclear in healthy cells and moves to the cytoplasm and out of the cell during certain death pathways, so a nuclear to cytoplasmic ratio or a measurement in the medium is the readout, not a whole-cell level.

**What controls does a cell death experiment need?**

A vehicle control from the same plate, a known inducer as a positive control, a viability measure, and a baseline death rate reported rather than hidden. Without those a difference cannot be attributed to the treatment.

**Why do my untreated cells show death markers?**

Ordinary culture stress: over-confluence, exhausted medium, a rough thaw or handling. Measure and report the baseline; a treatment that raises death from a high baseline is a much weaker result than one that raises it from a low one.

**How is a gasdermin D antibody result read?**

The same way as caspase: by the cleaved form. The full length protein sits in healthy cells and the pore-forming fragment is generated by cleavage, so a blot showing total gasdermin says only that the cell has the protein, while the cleaved band says the pathway ran.

## Cytoskeleton and matrix

Cytoskeletal antibodies are excellent reagents that fail for preparation reasons, and matrix antibodies report on material that survives the cell that made it. Both classes are also routinely over-interpreted: an intermediate filament that rises with a change of state gets reported as marking a cell type. This page covers the preparation and the interpretation.

1. **Choose fixation for the filament system**: Microtubules depolymerise in the cold and are poorly preserved by some fixatives, so an aldehyde protocol at temperature or a cold methanol protocol each preserve different aspects. A disappointing cytoskeletal image is nearly always fixation rather than antibody, and testing both on a known positive settles it in a day.
2. **Read an intermediate filament as a state, not a lineage**: Vimentin is expressed by mesenchymal cells and is induced in epithelial cells undergoing a change of state, so its presence is informative and its absence is not evidence of epithelial identity. Pair it with a lineage marker rather than using it alone.
3. **Treat matrix proteins as extracellular material**: Elastin and other matrix components are secreted, cross-linked and long lived, so their staining reports the tissue's history rather than current expression by nearby cells. Where synthesis is the question, transcript measurement or a pulse label answers it and a stain does not.
4. **Know that some targets are both intracellular and secreted**: Clusterin exists in secreted and intracellular forms with different biology, and antibodies differ in which they see. Read the immunogen and the application evidence, because a result about one form attributed to the other is a common error.
5. **Use the structure itself as the specificity control**: A cytoskeletal stain should look like the structure: filaments, not a haze. A diffuse cytoplasmic pattern where a filament network is expected is a warning sign whatever the intensity, and it is the fastest specificity check available.

**Why does my cytoskeletal stain look diffuse?**

Fixation, usually. Microtubules depolymerise in the cold and some fixatives preserve filaments badly. Test an aldehyde protocol at temperature against cold methanol on a known positive before changing antibody.

**Does an anti vimentin antibody identify mesenchymal cells?**

It identifies vimentin expression, which mesenchymal cells have and epithelial cells acquire during a change of state. Use it alongside a lineage marker rather than as an identity call on its own.

**What does an elastin antibody actually show?**

Deposited matrix, which is long lived and cross-linked. It reports the tissue's accumulated history rather than current synthesis by the cells beside it, which is a different question needing a different method.

**Should the antibody be validated in my tissue?**

For matrix targets particularly, yes. Fixation and cross-linking differ between tissues and archival material behaves differently again, so ask for images in a tissue and fixation like yours.

**Which form of clusterin does a clusterin antibody see?**

It depends on the reagent. Clusterin exists in secreted and intracellular forms with different biology, and antibodies differ in which they detect, so read the immunogen and the application evidence before buying. A result about one form attributed to the other is a common error.

## Labelling antibodies

Nucleotide incorporation is the only way to know which cells were in S phase during a defined window, and it is the stain most often reported as a failed antibody. The epitope is inside double stranded DNA and cannot be reached until the DNA is denatured, so a protocol that works for every other nuclear target reports nothing here. This page covers that and the damage markers used beside it.

1. **Denature the DNA, or the antibody cannot reach the epitope**: The incorporated nucleotide sits within double stranded DNA and is inaccessible. Acid treatment, heat or a nuclease step opens it, and without one the stain is blank. This single requirement accounts for most reported failures of these reagents.
2. **Accept what the denaturation costs you**: Acid denaturation damages protein epitopes and most fluorescent proteins, so a panel combining incorporation with surface or fluorescent protein markers has to be designed around it. Nuclease-based methods are gentler and are the route when other markers must survive.
3. **Fix the pulse length before the experiment**: The label marks cells that were synthesising during the pulse, so a short pulse measures a rate and a long one measures cumulative entry. Two studies with different pulse lengths are not comparable, and the length belongs in the methods.
4. **Run an unlabelled control through the whole protocol**: Cells that never saw the nucleotide, taken through denaturation and staining, show what the harsh protocol does to background. It is the control that matters most here and is routinely omitted.
5. **Read damage signalling markers as a modification, not a level**: Poly-ADP-ribose and related damage marks are generated transiently at damage sites, so the measurement is their appearance and their kinetics rather than a steady level. Include a known inducer and a time course, or the absence of signal means nothing.

**Why does my brdu antibody stain nothing?**

Almost certainly the DNA was not denatured. The epitope is inside double stranded DNA and needs acid, heat or a nuclease step to expose it, and no other nuclear stain requires this.

**Can I combine incorporation staining with surface markers?**

With care. Acid denaturation damages many protein epitopes and most fluorescent proteins, so either use a gentler nuclease-based method or validate that each marker survives the protocol you use.

**How long should the pulse be?**

It depends what you are measuring. A short pulse labels cells synthesising at that moment and reports a rate; a long pulse labels everything that entered S phase during it and reports cumulative entry. State the length in the methods.

**What does a par antibody report?**

Poly-ADP-ribose, generated transiently at sites of DNA damage. It is a kinetic readout rather than a level, so an experiment needs a known inducer and a time course for a result to be interpretable.

## Stem and epithelial markers

Some targets have no good antibody, and saying so is more useful than selling one. Low abundance receptors on rare cells are the classic case: the protein is scarce, the cells are few, and a reagent that gives a plausible pattern is very hard to disprove. This page covers four epithelial and stem-associated targets, what each one needs, and when to stop looking for an antibody.

1. **Accept that some receptors have no reliable antibody**: Low abundance surface receptors on rare cell populations are the hardest antibody targets there are, and several widely cited ones have a poor track record in independent hands. Before committing a study, search for independent validation rather than catalog images, and treat a plausible pattern as a hypothesis.
2. **Use a reporter line where the antibody is unreliable**: A knock-in reporter at the endogenous locus reports expression without a reagent at all, and for the hardest of these targets it is the standard the field actually uses. It costs an editing project and it is the honest route when the antibody question has no good answer.
3. **Match the preparation to a secreted mucin**: Heavily glycosylated secreted proteins run as smears, are lost in some fixations and are extracted poorly by standard lysis. Expect to test fixation and extraction rather than assuming a protocol carries over from a cytoplasmic protein.
4. **Check the epitope region for surface work**: For a surface target being stained on intact cells or used to sort them, the epitope must be extracellular. Many catalog reagents against these targets are raised against intracellular regions, which is stated on the datasheet and is the commonest reason a sort fails.
5. **Validate on tissue where the expected pattern is known**: These markers have well described distributions in normal tissue, which gives a built-in control: the right cells in the right place is evidence, and a uniform stain across a tissue that should be restricted is not.

**Why is an lgr5 antibody so hard to use?**

The protein is at low abundance on a small population, which is the hardest combination there is. Several widely sold reagents have a poor record in independent hands, and a plausible-looking pattern is very hard to disprove without a genetic control.

**What should I use instead?**

A knock-in reporter at the endogenous locus, which reports expression without a reagent, or a transcript-level readout. For the hardest targets this is what the field relies on rather than staining.

**Why does my muc2 antibody give a smear?**

Heavy glycosylation and secretion. Mucins run broadly and high, are lost in some fixations and extract poorly under standard lysis, so both the fixation and the extraction need testing rather than inheriting.

**Can I sort cells with a nectin-4 antibody?**

Only with an extracellular epitope and a validated live-cell protocol. Check the immunogen region first, because many reagents for these targets are against intracellular sequences and cannot stain an intact cell.

## Drug metabolism antibodies

The cytochrome P450 family is the hardest case for antibody specificity in routine use, because its members share most of their sequence and differ in the regions that matter. An antibody raised against a conserved stretch will report the whole family, and nothing about the band size will tell you. This page covers what evidence to demand and when an activity measurement is the better experiment.

1. **Demand cross-reactivity data against the family, by name**: The useful evidence is a panel of recombinant family members blotted side by side, showing the antibody detects one and not the others. A datasheet naming only the intended target has not addressed the question that matters for this family.
2. **Prepare microsomes rather than whole lysate**: These enzymes sit in the endoplasmic reticulum membrane, so a microsomal preparation concentrates them substantially against soluble protein. It is a centrifugation step and it usually makes more difference than any change of antibody.
3. **Do not over-denature a membrane enzyme**: As with other membrane proteins, hard boiling promotes aggregation. A lower denaturation temperature is worth testing before concluding the reagent does not work, and a lane that shows material stuck in the well is the diagnostic.
4. **Use recombinant enzyme as the positive and a null system as the negative**: Single-enzyme recombinant preparations are widely available and make an unambiguous positive control. A system expressing none of the family is the corresponding negative, and together they settle specificity far better than tissue alone.
5. **Consider whether activity is the real question**: A selective probe substrate measured by chromatography reports what the enzyme does, which is usually what a study needs, and it distinguishes family members by their catalytic selectivity rather than by an epitope. Protein level and activity frequently diverge.

**How do I know a cyp2c9 antibody is specific?**

By a blot of recombinant family members side by side showing detection of the target and not its close relatives. Within this family, sequence identity is high enough that specificity has to be demonstrated rather than assumed.

**Why is my CYP2E1 antibody signal weak in liver lysate?**

Whole lysate dilutes a membrane enzyme with abundant soluble protein. Prepare microsomes, and check the denaturation temperature, because these enzymes aggregate when boiled hard.

**Should I measure protein or activity?**

Activity, usually. A selective probe substrate measured by chromatography reports function and distinguishes family members catalytically. Protein level and activity diverge often enough that the choice matters.

**Can these antibodies be used for immunohistochemistry?**

Some are validated for it, and the same cross-reactivity question applies with less ability to check by size. Ask for staining in a tissue with a known expression pattern and, ideally, alongside a null control.

**Is a protein blot evidence of enzyme activity?**

No. Activity depends on localisation, cofactors, modification and inhibitors as well as on how much protein is present. If the claim concerns flux, measure activity or a product and use the blot as supporting evidence.

## Growth factor receptors

Receptor signalling blots go wrong in the lysis buffer, not on the membrane. Phosphorylation is reversed within seconds of lysis unless inhibitors are present from the start, so a phospho blot with no signal is usually a handling result rather than a biological one. This page covers the difference between total and phospho reagents and the controls a signalling experiment needs.

1. **Decide whether the question is amount or activity**: A total receptor antibody reports how much is present, which changes over hours to days. A phosphorylation-specific antibody reports activation at a residue, which changes in seconds to minutes. They answer different questions and a study usually needs both, on the same membrane.
2. **Put phosphatase inhibitors in the buffer before you lyse**: Phosphatases are active the moment a cell breaks. Inhibitors go into the lysis buffer fresh, the plate goes on ice, and the lysate is handled cold and quickly. Most missing phospho signal is lost in the first minute, and no amount of antibody optimisation recovers it.
3. **Include a stimulated positive and an inhibited negative**: Ligand-stimulated cells give a signal the antibody must detect, and cells treated with a selective inhibitor give a signal it must lose. That pair demonstrates both the reagent and the experiment, and a blot without them cannot show much.
4. **Normalise phospho signal to total, not to a housekeeping protein**: The informative quantity is the fraction of receptor that is phosphorylated, so the phospho signal is read against total receptor on the same samples. Normalising to a loading control alone confuses a change in activation with a change in abundance.
5. **Check the residue and the species the antibody was raised against**: Phospho antibodies are specific to a residue, and different residues report different downstream consequences. The residue number should be in your methods, and the sequence around it has to be conserved in your species for the reagent to work.

**Why is there no signal from my phospho antibody?**

Almost always sample handling. Phosphatases act within seconds of lysis, so inhibitors have to be in the buffer before lysis and everything kept cold and fast. A stimulated positive control tells you immediately whether the problem is handling or biology.

**Should phospho signal be normalised to total receptor or to a loading control?**

To total receptor on the same samples. The meaningful quantity is the fraction activated; normalising only to a housekeeping protein cannot separate more activation from more receptor.

**Does the residue matter?**

Yes. Different phosphorylation sites recruit different partners and report different downstream events, so the residue belongs in the methods and a result about one site should not be generalised to the receptor being active.

**Can I strip and reprobe for total after phospho?**

Commonly done and it costs signal and risks uneven stripping. Running duplicate membranes, or a fluorescent two-color detection on one membrane, is more reliable where the ratio is the measurement.

**Which reagent do I need for a signalling experiment?**

Both: a modification-specific antibody and the matching total-protein antibody, read as a ratio, with a stimulated positive and an inhibited negative in the same run. A phospho blot alone cannot distinguish more signalling from more receptor.

## Tumour suppressor antibodies

Tumour suppressors are read for absence as often as for presence, and an antibody asked to demonstrate that a protein is missing has to be held to a much higher standard than one showing it is there. Add that two of these three are informative only as a modification state or as a focus count, and the class needs its own set of controls. This page sets them out.

1. **Read RB1 as a phosphorylation state, not an amount**: The protein is active when hypophosphorylated and inactivated by phosphorylation through the cycle, so a total antibody reports presence while a phospho-specific reagent reports the state that matters. Say which you used and at which residue, because the two support different conclusions.
2. **Treat loss of expression as the hardest claim to make**: Showing a protein is absent requires an antibody that would certainly have detected it, which means a positive control on the same slide or membrane and, ideally, a graded series. An absent band on a blot with no positive lane demonstrates nothing about the sample.
3. **Count DNA damage markers as foci, not as intensity**: Damage response proteins concentrate into discrete nuclear foci at break sites, so the readout is foci per nucleus over time. Total nuclear intensity averages the signal away and is the commonest way a damage experiment reports nothing.
4. **Include an inducer and a time course**: A known damaging treatment produces foci that appear and then resolve, which demonstrates the assay works and gives the kinetics that are usually the interesting result. Without an inducer, an absence of foci is uninterpretable.
5. **Use an isogenic null where the biology allows**: For loss-of-function claims, a line known to lack the protein run alongside is the strongest available control, and for these genes such lines are widely available. It settles in one lane what an isotype control cannot.

**Total or phospho-specific for RB1?**

Phospho-specific if the question is about the protein's activity, because it is inactivated by phosphorylation rather than by degradation. A total antibody answers whether the protein is present, which is a different and usually easier question.

**How do I show a tumour suppressor is lost, for example with a CDKN2A antibody?**

With a positive control that would certainly have been detected, on the same membrane or slide, and preferably an isogenic null as a negative. Absence of signal in isolation is not evidence of absence of protein.

**Why does my 53BP1 antibody stain look like nothing happened?**

Almost always because the readout was total nuclear intensity rather than foci. These proteins concentrate at break sites, so count discrete foci per nucleus and run a time course after a known inducer.

**Does a transcript measurement substitute?**

Not for these. Loss frequently happens by deletion, by promoter methylation or by protein destabilisation, and transcript and protein disagree often enough that the protein measurement is the one that supports the claim.

## Autophagy and stress

Autophagy readouts are the most commonly misread in cell biology, because every marker in the pathway can rise either because more is being made or because less is being degraded. Without a step that blocks degradation, the two are indistinguishable and the conclusion is a coin toss. This page covers what each marker reports and the control that makes any of it interpretable.

1. **Know which compartment each marker is in**: A lysosomal membrane protein marks the lysosome, not the autophagosome, and its colocalisation with an autophagosome marker is what reports fusion. Treating a lysosomal marker as an autophagy readout confuses the destination with the process.
2. **Block degradation, or the result is ambiguous**: Every marker downstream of autophagosome formation accumulates when degradation is blocked. The experiment therefore runs with and without a lysosomal inhibitor, and the difference between the two arms is flux. An experiment with one arm cannot distinguish more autophagy from less degradation.
3. **Read cargo receptors as turnover**: Receptors that link cargo to the autophagosome are themselves degraded, so their level falls when autophagy runs and rises when it is blocked. That inversion catches people out and is exactly why the inhibitor arm matters.
4. **Count granules and puncta, not intensity**: Stress granule and autophagosome markers concentrate into discrete structures, so the readout is number and size per cell over time. Total intensity averages the structure away and reports very little.
5. **Include a known inducer and a known inhibitor**: Starvation or a pharmacological inducer produces the expected change, and an inhibitor removes it. Those two arms demonstrate that the assay detects the pathway at all, which a treated-and-untreated pair does not.

**Does a LAMP-2 antibody measure autophagy?**

No. It marks the lysosome, which is where autophagosomes deliver cargo. Its colocalisation with an autophagosome marker reports fusion; on its own it reports lysosomal abundance.

**Why do I need a lysosomal inhibitor arm?**

Because every downstream marker accumulates both when more autophagy happens and when degradation is blocked. The difference between the inhibited and uninhibited arms is flux, and a single arm cannot distinguish the two causes.

**Why did my NBR1 antibody signal go up when I induced autophagy?**

Cargo receptors are degraded along with their cargo, so their level falls with active autophagy and rises when it is blocked. If it rose, either the pathway is blocked or the inducer did not work, and the control arms tell you which.

**How should stress granules stained with a G3BP1 antibody be quantified?**

As number and size per cell, with a time course after a known stressor. They form and dissolve on a timescale of minutes to tens of minutes, so a single time point can easily sit before or after the peak.

## Mitochondrial markers

Mitochondrial markers are used to normalise, to fractionate and to claim a change in mitochondrial content, and only the first two of those are straightforward. A marker protein can change independently of organelle mass, and a fraction is an enrichment rather than a purification. This page covers what each marker is good for and the controls that keep the conclusion honest.

1. **Use a matrix enzyme as a mass marker, with care**: Citrate synthase is a matrix enzyme conventionally used as a proxy for mitochondrial content because it is relatively stable across conditions. It is a proxy, not a measurement, and it can itself be regulated, so a change in it is a hypothesis about organelle mass rather than a demonstration.
2. **Use a respiratory chain subunit to ask a different question**: A subunit of the respiratory complexes reports on the chain itself, which can change independently of matrix content under stress or in disease. Blotting one of each is far more informative than either alone.
3. **Always run the cytosolic counterpart**: A cytosolic enzyme on the same membrane shows whether a mitochondrial fraction is contaminated and whether a cytosolic fraction is contaminated the other way. A fractionation shown with only its own marker is an unverified claim.
4. **Report enrichment, not purity**: Differential centrifugation enriches; it does not purify, and adjacent membranes co-sediment. Blot markers for the endoplasmic reticulum and the nucleus too, and report the enrichment factor rather than implying a clean fraction.
5. **Pair protein markers with a functional or copy number measure**: Mitochondrial DNA copy number and respiration measurements answer the content and function questions that a marker blot only approximates. Where the conclusion is about mitochondrial biogenesis, at least two independent measures are expected.

**Is a citrate synthetase antibody a measure of mitochondrial content?**

It is the conventional proxy, because the enzyme is comparatively stable across conditions. It is not a direct measurement and it can itself be regulated, so a change in it supports a hypothesis about content rather than proving one.

**Why blot with a COXIV antibody as well?**

Because the chain can change independently of matrix content, particularly under stress or in mitochondrial disease. One marker from each compartment gives a far more complete picture than either alone.

**What does the cytosolic LDH antibody lane tell me?**

Whether the fractions are cross-contaminated. A mitochondrial fraction carrying a strong cytosolic signal is not the fraction the experiment assumed, and a cytosolic fraction carrying mitochondrial signal is equally informative.

**Can I claim more mitochondria from a marker blot alone?**

Not convincingly. Pair it with mitochondrial DNA copy number or a respiration measurement. Claims about biogenesis are expected to rest on more than one kind of evidence.

## Neurodevelopmental markers

Developmental markers are stage specific, which makes the control harder than the stain: a negative result can mean the protein is absent, or that the tissue is at the wrong stage, or that the retrieval was wrong. This page covers the nuclear factors and the mature neuronal markers used beside them, and the controls that distinguish those three explanations.

1. **Use a stage-matched positive control, not just a positive tissue**: These factors are expressed in defined regions at defined stages. A positive control from the wrong stage is not a control, and it is the commonest reason a developmental stain is reported as failed when the protocol was fine.
2. **Treat nuclear factors as low abundance nuclear targets**: They need a retrieval and a permeabilisation matched to the nucleus, and frequently an amplified detection. A protocol that works for a cytoskeletal or synaptic marker will report absence for these.
3. **Read a synaptic marker as puncta**: Presynaptic markers concentrate at terminals, so the readout is puncta number and density along a process rather than total intensity. Total intensity conflates more synapses with more protein per synapse.
4. **Know what a neuronal cytoplasmic marker does and does not count**: A pan-neuronal cytoplasmic marker fills the cell and is excellent for morphology and for confirming neuronal identity, and it is a poor basis for counting because processes overlap. Count on a nuclear marker and describe with a cytoplasmic one.
5. **Fix the fixation and record the interval**: Several of these epitopes degrade with time to fixation and behave differently in over-fixed archival blocks. Record fixation time and post-mortem interval with every sample; a cohort inconsistent in these cannot be compared.

**Why is my developmental marker negative?**

It may be the wrong stage rather than a failed stain. These factors are expressed in defined regions at defined times, so a stage-matched positive control is what separates absent protein from a protocol problem.

**What retrieval do an NKX2.2 antibody and an MECP2 antibody need?**

Heat induced retrieval, tested at both a citrate buffer near pH six and a Tris buffer near pH nine on a known positive. The right condition differs between these targets and cannot be assumed from a related one.

**How should a synapsin antibody stain be quantified?**

As puncta density along a defined length of process or within a defined region, with fixed acquisition settings. Total intensity cannot separate more synapses from more protein per synapse.

**Can I count neurons with a UCHL1 antibody or another cytoplasmic pan-neuronal marker?**

Poorly, because processes overlap and cell boundaries are unclear. Count on a nuclear neuronal marker and use the cytoplasmic one to describe morphology and confirm identity.

## Kinase pathway antibodies

Pathway antibodies are bought in pairs and used as if they were one reagent. A total antibody reports how much kinase is present, which moves over hours; a phospho reagent reports activity at a residue, which moves in seconds and is lost in the lysis buffer if inhibitors were not already there. This page covers the pairs, the handling and the one control that settles a doubtful band.

1. **Buy the pair and read the ratio**: Phospho signal is read against total protein on the same samples, because the informative quantity is the fraction activated. Normalising a phospho band to a loading control alone cannot separate more activation from more kinase.
2. **Put phosphatase inhibitors in before you lyse**: Dephosphorylation begins the moment a cell breaks. Inhibitors go into the buffer fresh, the plate goes on ice and the lysate is handled cold and fast. Most missing phospho signal is lost in the first minute and no optimisation recovers it.
3. **Use a selective inhibitor as the specificity control**: A band that disappears when the kinase is inhibited is the kinase's band. This is the strongest and cheapest specificity evidence available for a phospho reagent and it is routinely skipped in favor of an isotype control that proves much less.
4. **Check the residue and its conservation**: Phospho antibodies are residue specific and different sites report different downstream consequences. Put the residue in the methods, and confirm the surrounding sequence is conserved in your species before assuming cross-reactivity.
5. **Remember that some of these move rather than change**: A transcriptional co-activator is regulated by shuttling between cytoplasm and nucleus, so a whole-cell blot can be flat while the biology changes entirely. For those targets, fractionate or measure a per-cell nuclear ratio by imaging.

**Why is my phospho mTOR antibody band missing?**

Handling, almost always. Phosphatase inhibitors must be in the lysis buffer before lysis, and everything kept cold and fast. A stimulated positive control tells you in one lane whether it is handling or biology.

**What is the best specificity control for a phospho antibody?**

A selective inhibitor of the kinase. A band that disappears under inhibition is the kinase's band, which is far stronger evidence than an isotype control and usually cheaper than a genetic approach.

**Should phospho be normalised to total or to a loading control?**

To total on the same samples. The meaningful quantity is the fraction of the protein that is phosphorylated, and a loading control alone conflates activation with abundance.

**Why is my YAP1 antibody blot flat when the treatment clearly works?**

Because it is regulated by moving rather than by amount. Fractionate into nuclear and cytoplasmic pools, or image and measure a nuclear to cytoplasmic ratio per cell.

**Is a scaffold protein a safe loading control?**

Only after you have shown it does not change under your treatment and that it is loaded inside the linear range. Total protein staining measures what was loaded and avoids the assumption altogether.

## Epithelial and junction markers

Epithelial markers give you a specificity control for free, because the staining pattern is the answer: a junction protein should draw a line between cells and a keratin should fill a cytoplasm. A diffuse haze in either case is a failed stain regardless of intensity. This page covers what each marker identifies and how to read the pattern.

1. **Know that keratin subtypes are informative and pan-keratin is not**: A specific keratin marks a subset of epithelia and is used to distinguish lineages and ducts; a pan-keratin cocktail confirms epithelial identity and no more. Choose from the question, and state which reagent produced a result.
2. **Recognize that E-cadherin and CDH1 are one protein**: They are the protein name and the gene name for the same target, and catalog entries use both. Reagents against different domains behave differently after fixation, so compare on the epitope rather than on the naming.
3. **Read a tight junction protein as a continuous line**: A junctional protein should appear as a continuous line at cell borders in an intact monolayer, and breaks in that line are the informative result in barrier experiments. Cytoplasmic staining of a junction protein means either disassembly or a failed stain, and the control tells you which.
4. **Use the pattern as the specificity control**: Membrane and junctional proteins have unmistakable patterns. A stain that is the right intensity in the wrong pattern is a warning that costs nothing to notice and is the fastest check available on these targets.
5. **Match fixation to the junction**: Junctional epitopes are sensitive to fixation and to permeabilisation, and methanol and aldehyde protocols preserve them differently. Test both on a confluent monolayer known to have intact junctions before interpreting anything.

**Is an E-cadherin antibody different from a CDH1 antibody?**

No, they are the protein and gene names for the same molecule, and suppliers list both. What differs between products is the epitope domain, which changes how they behave after fixation and whether they work on intact cells.

**Why is my junction protein stain cytoplasmic?**

Either the junctions have genuinely disassembled, which is often the result you are looking for, or the fixation and permeabilisation are wrong. A confluent untreated monolayer stained alongside distinguishes the two immediately.

**Should I use a specific keratin or a pan-keratin cocktail?**

A specific keratin when the question is which epithelial lineage or which duct compartment. A pan-keratin cocktail when the question is only whether the cells are epithelial at all.

**What is the quickest specificity check on these markers?**

The pattern. A junction protein should draw a line at cell borders and a keratin should fill a cytoplasm. Right intensity in the wrong pattern is a failed stain whatever the datasheet says.

## Immune receptor antibodies

Chemokine and cytokine receptors are among the hardest antibody targets in immunology: multipass membrane proteins at low abundance whose extracellular loops are short. They blot badly and stain intact cells well, which is why nearly all reliable data on them comes from cytometry. This page covers that and the receptor and ligand pairs that have to be measured together.

1. **Measure surface receptors on intact cells**: Multipass receptors aggregate when boiled, run as smears and are frequently invisible on a blot at native abundance. Flow cytometry with an antibody against an extracellular loop is the measurement these targets support, and it also separates surface from internalised pools.
2. **Check the epitope is extracellular before buying**: Short extracellular loops make good extracellular reagents scarce, and many catalog antibodies against these targets are raised against intracellular sequences. The immunogen region is on the datasheet and it decides whether a surface stain is possible at all.
3. **Measure a receptor and its ligand together**: Where a soluble receptor exists it acts as a decoy and changes what a ligand measurement means, so measuring one without the other gives a number that cannot be interpreted. Plan the pair from the start.
4. **Use a null population as the negative**: Immune samples contain populations known not to express a given receptor, which gives an internal negative in the same tube. That is a better control than an isotype and it is free.
5. **Titrate on the real sample type**: Receptor density is low and the separation between positive and negative is small, so titration matters more here than on an abundant marker. Choose the concentration giving the best separation, not the strongest signal.

**Why can I not see a chemokine receptor on a western blot?**

They are multipass membrane proteins at low abundance that aggregate when boiled and run as smears. Cytometry on intact cells with an extracellular-epitope antibody is the measurement these targets actually support.

**How do I know the antibody will stain intact cells?**

Check the immunogen region. Only an antibody against an extracellular loop can bind an intact cell, and for receptors with short loops those reagents are scarcer and worth identifying before ordering.

**Why measure receptor and ligand together, with an ST2 antibody and an IL-33 antibody?**

Because a soluble form of the receptor acts as a decoy for the ligand, so a ligand concentration measured alone can be misleading. The pair has to be interpreted together.

**What is the best negative control in an immune sample?**

A population in the same tube known not to express the receptor. It controls for the staining, the instrument and the sample at once, which no isotype control can do.

## Hypoxia and metabolic

The hypoxia transcription factor is the clearest case in cell biology of a result decided before the gel: the protein is degraded within minutes of a cell meeting oxygen, so a sample taken out of the chamber and lysed on the bench reports very little. This page covers that handling, the transporters measured beside it, and the controls each needs.

1. **Lyse inside the chamber, or accept the loss**: The hypoxia factor is hydroxylated and degraded rapidly once oxygen is available, so plates taken to a bench lose most of the signal before lysis. Lysing inside the hypoxia chamber, or working fast on ice with the lysis buffer already prepared, is the difference between a band and nothing.
2. **Use a proteasome inhibitor arm as the positive control**: Blocking degradation stabilises the factor and gives a strong band that shows the antibody and the blot work. It also proves that a missing band in the experimental arm is biology rather than handling.
3. **Know that a nuclear fraction reports the active pool**: The factor acts in the nucleus and the whole-cell level includes protein on its way to destruction. Fractionating, or imaging a nuclear signal per cell, reports the pool that is doing something.
4. **Treat the transporters as multipass membrane proteins**: Amino acid and other solute transporters aggregate when boiled and need a detergent that keeps the epitope. Denature at a lower temperature, prepare membranes rather than whole lysate, and use a genetic negative because background is high.
5. **Use a modification-aware reagent where the target is a modifier**: Enzymes that add a sugar or another modification are read through their substrates as much as through their own level, so a total antibody tells you the enzyme is present while a substrate-modification antibody tells you it is working.

**Why does my hif antibody show no band from hypoxic cells?**

The protein is degraded within minutes of reoxygenation, so it is usually lost between the chamber and the bench. Lyse inside the chamber, or work fast and cold, and run a proteasome inhibitor arm as a positive control.

**What proves the blot would have worked?**

A proteasome inhibitor arm. Blocking degradation stabilises the protein and gives a strong band, which shows the reagent and the transfer are fine and that the experimental result is biology.

**Why do my SLC1A5 antibody and SLC7A11 antibody blots look like nothing?**

Multipass transporters aggregate when boiled and stay in the well. Denature at thirty seven or seventy degrees, prepare membranes rather than whole lysate, and use a knockdown as the specificity control.

**Should the hypoxia factor be measured in the nucleus?**

Where the question is about activity, yes. The whole-cell pool includes protein destined for degradation, so a nuclear fraction or a per-cell nuclear intensity is a better proxy for the active pool.

**Why does my membrane enzyme vanish from the gel?**

Standard lysis and heating aggregate multi-pass membrane proteins so that they never enter the gel. Use the gentler solubilisation the validation used, avoid boiling, and enrich the membrane fraction.

## Neuropeptides and growth factors

Secreted peptides and growth factors are made in one place and act in another, so where you measure decides what you learn. They also exist as precursors that are processed, and antibodies differ in which form they see, which explains a great deal of the disagreement between published concentrations. This page covers both problems.

1. **Measure where the molecule acts, not only where it is made**: A lysate tells you a cell contains the peptide; a fluid or a conditioned medium tells you what was released. For most questions about a secreted factor the fluid is the sample, and a tissue stain answers a different question about the source.
2. **Establish whether the reagent sees precursor or mature peptide**: These molecules are made as precursors and processed, and the biologically active species is usually the processed one. An antibody against a precursor region reports something different, and this is the single largest source of disagreement between assays.
3. **Match the assay's matrix and qualify it in yours**: An assay validated in serum used on cerebrospinal fluid, conditioned medium or tissue extract is unvalidated until spike recovery and dilution linearity are demonstrated in that matrix.
4. **Fix the pre-analytical handling and record it**: Peptides are proteolysed rapidly in blood, so tube type, protease inhibitors, time to centrifugation and freeze-thaw all change the number. Fix the protocol before the first sample and apply it to every one.
5. **Use a matched pair for immunoassay and buy it as a pair**: Capture and detection antibodies must see different epitopes without blocking each other. For small peptides the available epitopes are few, which is exactly why an assembled pair rarely works and a validated one is worth paying for.

**Should a secreted peptide be measured in lysate or in fluid?**

In fluid for almost every question about signalling, because the molecule acts where it is released. A lysate or a tissue stain tells you about the source cell, which is a different question worth asking separately.

**Why do two assays using a gastrin antibody or a BDNF antibody disagree?**

Usually because they see different forms. These peptides are processed from precursors, and an antibody against a precursor region reports a different species from one against the mature peptide.

**What handling do peptide samples need?**

Protease inhibitors in the tube where the assay requires them, rapid centrifugation, cold storage and minimal freeze-thaw. Peptides degrade far faster than proteins in blood and the number moves accordingly.

**Can I build my own immunoassay for a small peptide?**

It is much harder than for a protein, because a short peptide offers few epitopes for a capture and detection pair that do not block each other. Buying a validated pair is usually the cheaper route by the end.

## LIMS database

Laboratories ask for database access and usually mean one of two very different things: reading their own data for analysis, which is reasonable and useful, or writing to it directly, which bypasses every control the system exists to provide. This page separates them and covers the questions worth asking a vendor about what lies underneath.

1. **Ask for read access, and expect it through a view or an API**: A read-only view or a documented interface gives you your data for analysis without exposing the internal schema you would then depend on. A vendor unwilling to provide any read route is a vendor whose data you cannot analyze, which is a real cost over a decade.
2. **Never write directly to the tables**: Direct writes bypass validation, the audit trail and any business logic the application enforces, which means the record no longer reflects what the system believes happened. In a regulated laboratory that is a data integrity finding; in any laboratory it is a corruption waiting to be discovered.
3. **Establish how the audit trail is stored and whether it can be read**: The audit trail is the part of the record an inspector reads, and its usefulness depends on whether it can be queried rather than only displayed a row at a time. Ask to see an audit trail extract for a real change.
4. **Ask what happens to old data on upgrade and on archive**: Schemas change between major versions, and archived data that can only be read by a retired version is data you have lost. Ask specifically how historical records are migrated and whether an archive remains queryable.
5. **Confirm the export before you sign**: A full export including the audit trail in an open format, tested during the evaluation, is the single most useful hour of a software trial. It decides whether you can ever leave, and therefore what every renewal conversation looks like.

**Should I have direct access to the lims database?**

Read access, yes, ideally through a documented view or interface. Write access, no: direct writes bypass validation, business logic and the audit trail, and leave a record that does not describe what the system thinks happened.

**Who owns the schema?**

The vendor, in practice, and it changes between versions. That is why analysis should run against a stable view or an interface rather than against the internal tables, which can be reshaped by an upgrade without warning.

**What does the audit trail have to do?**

Record who changed what, when and why, be protected from alteration, and be readable. Under 21 CFR Part 11 it also has to be reviewed rather than merely enabled, which means it has to be queryable in practice.

**How do I get my data out at the end?**

Test the export during the trial, including the audit trail, in an open format. An export that turns out to be a set of screen reports is not an export, and discovering that at the end of a contract is the worst moment to discover it.

## What a LIMS changes

Most published lists of LIMS benefits are written by people selling one. The honest version is shorter, and it separates what the software does by existing from what it only does if the laboratory configures it and uses it. This page is that separation, because a business case built on the second kind without planning for adoption is the reason LIMS projects disappoint.

1. **Transcription disappears where instruments are interfaced**: The largest and most reliable gain is removing manual transfer of results from an instrument to a sheet to a report. It is measurable in hours and in errors, and it happens only for the instruments actually interfaced, which is why the interface list is the business case.
2. **Samples become findable**: Position-level location, status and history mean nobody searches a freezer with the door open or asks three people where a sample went. This is worth more in a busy laboratory than most feature comparisons suggest and it requires that people actually scan.
3. **Traceability stops being a project**: Which lot, which instrument, which analyst, which method version produced a result becomes a query rather than a week of paper. For accredited and regulated laboratories this is the difference between an audit that takes a day and one that takes a fortnight.
4. **Reporting becomes reproducible**: A certificate generated from approved results is identical every time it is produced and can be regenerated years later. Reports assembled by hand cannot make that promise, and in a dispute the promise is the point.
5. **Capacity and backlog become visible**: Turnaround by test, backlog by analyst and where work is stuck are aggregates the system has and a spreadsheet does not. Whether anyone acts on them is a management question, not a software one.

**What are the real advantages of LIMS system adoption?**

Transcription removed where instruments are interfaced, samples findable to a position, traceability as a query rather than a paper chase, reproducible reporting, and visible capacity. The first two are the ones laboratories feel immediately.

**Which benefits depend on us rather than on the software?**

All of them to some degree, and adoption most of all. A system nobody enters data into promptly is worse than the spreadsheet it replaced, and the commonest cause of a disappointing implementation is configuration nobody was consulted on.

**What will a LIMS not fix?**

A workflow nobody agrees on, a method that is not written down, or a backlog caused by capacity. It makes those things visible, which is useful and is not the same as solving them.

**How long before the benefit appears?**

Sample registration and location pay back within weeks of go-live. Instrument interfaces pay back as each one lands. Reporting and traceability pay back at the first audit or the first dispute, which is when the business case is actually tested.

## sgRNA synthesis

A guide RNA can be made two ways and the choice changes the experiment. Chemical synthesis gives a defined, modifiable molecule at a length that is now routine; in vitro transcription is cheaper at scale and leaves a triphosphate end that cells notice. This page covers the routes, the modifications and the purity grade that decide editing efficiency.

1. **Choose chemical synthesis where modifications matter**: Synthetic guides can carry modifications at defined positions, which is what raises stability and lowers the innate immune response in primary cells. For a guide that will go into cells that matter, this is usually the deciding factor rather than price.
2. **Use transcription for scale and for length**: In vitro transcription is cheaper per guide at volume and handles longer molecules easily, which suits screening libraries and early optimisation. The transcript carries a triphosphate end that can trigger an innate response, and treatment to remove it is worth specifying.
3. **Specify the modifications by position, not by name**: End modifications on the first and last few nucleotides are the common pattern and they materially raise editing in primary cells. Ask which positions are modified and with what, because the supplier's shorthand names differ.
4. **Buy the purity grade the application needs**: Truncated products from incomplete synthesis compete for the nuclease without cutting, so purification grade affects editing efficiency directly. Screening tolerates a lower grade; a single experiment in precious primary cells does not.
5. **Check the certificate carries identity and purity**: Mass spectrometry confirming the expected mass and a purity figure by a named method are the minimum. A concentration alone is a shipping note, and for a guide it tells you nothing about the fraction that is full length.

**Synthetic or transcribed sgRNA?**

Synthetic where modifications, defined ends and consistency matter, which covers most work in primary cells. Transcribed where cost per guide at scale matters, such as screening, accepting the triphosphate end and its innate immune consequences.

**Do modifications really improve editing?**

In primary cells and in vivo they make a substantial difference, by raising stability and reducing the innate response to a foreign RNA. In robust cell lines the difference is much smaller, which is why the answer depends on the destination.

**What purity grade should I order?**

The highest you can justify for experiments that matter, because truncated guides occupy the nuclease without cutting. For large screens a lower grade is the practical trade, and it should be a decision rather than a default.

**What should the certificate show?**

Identity by mass spectrometry against the expected mass, purity by a named method with a number, and concentration with the method used. Those three make the material traceable; a concentration alone does not.

## XRF analysers

X-ray fluorescence measures elemental composition without dissolving anything, which makes it uniquely convenient and uniquely easy to over-trust. It is a surface technique, its calibration is matrix specific, and it cannot see the lightest elements at all. This page covers what it is genuinely good for and where a digestion and a plasma instrument remain the answer.

1. **Understand what it can and cannot detect**: The technique works from roughly sodium upwards, with sensitivity improving for heavier elements. Carbon, nitrogen, oxygen and the lightest metals are outside its reach entirely, and no configuration changes that. If your analyte is light, this is the wrong instrument.
2. **Treat it as a surface measurement**: The X-rays penetrate a limited depth, so the result describes a surface layer rather than the bulk. Coatings, corrosion, contamination and inhomogeneity all bias it, which is why sample presentation matters more here than with a digestion method.
3. **Calibrate against a matrix like your sample**: Fluorescence intensity depends on the whole matrix, not just on the element, so a calibration built on alloys will not describe a polymer or a soil. Matrix-matched certified reference materials are what make a result quantitative rather than indicative.
4. **Decide handheld against benchtop honestly**: A handheld instrument is excellent for sorting, screening and field identification. A benchtop instrument in a controlled geometry with prepared samples is what produces numbers you would defend. Buying one and expecting the other is the usual disappointment.
5. **Prepare the sample if the number matters**: Grinding, pressing a pellet or fusing a bead removes the surface and inhomogeneity problems and turns a screening technique into a quantitative one. That preparation is most of the work and it is what separates the two use cases.

**What can an xrf analyzer not detect?**

The light elements. It works from around sodium upwards, so carbon, nitrogen, oxygen and the lightest metals are outside its range entirely, whatever the configuration.

**Is a handheld XRF quantitative?**

For sorting and screening in a known material family, it is fast and reliable. For a number you would defend, prepared samples in a controlled benchtop geometry with a matrix-matched calibration is the honest answer.

**Why does my result differ from a laboratory analysis?**

Usually the surface and the matrix. XRF reads a thin surface layer and its response depends on the whole matrix, so a coating, corrosion or a calibration built on a different material all shift it.

**Does it need a radiation licence?**

It produces X-rays and its use is regulated, with registration, training and monitoring requirements that vary by jurisdiction. Settle that before purchase rather than after delivery.

**Which elemental technique should I choose?**

Fluorescence for fast, non-destructive screening at parts per million; dissolution-based absorption or plasma techniques for trace quantification at much lower levels, with plasma instruments covering many elements at once.

## Endotoxin testing

Endotoxin and bioburden are tested together and measure different things: one is a pyrogenic contaminant from bacterial cell walls that survives sterilisation, the other is the count of viable organisms present. Both are release tests for parenteral products and both are commonly ordered without the product-specific validation that has to come first.

1. **Do the inhibition and enhancement study first**: Before any routine testing, the laboratory has to show your product does not interfere with the assay, by spiking a known amount of endotoxin into it and recovering it within the required window. That validation is product-specific and it takes time nobody schedules.
2. **Fix the dilution and the endotoxin limit**: The limit follows from the dose and the route of administration, and the maximum valid dilution follows from the limit and the assay sensitivity. Those two numbers define the test, and they belong in the request rather than being derived by the laboratory from assumptions.
3. **Understand what bioburden counts**: Bioburden is the viable organism count before sterilisation and it drives the sterilisation process design. It is not a sterility test and it is not endotoxin; a product can be low bioburden and high endotoxin, because killing the organisms leaves their cell wall material behind.
4. **Use depyrogenated glassware and controls throughout**: Endotoxin is everywhere and survives autoclaving, so glassware is depyrogenated by dry heat and every run carries a positive product control. A negative result from a contaminated workflow is not a negative result.
5. **Ask what the report will state**: The method, the sample dilution, the endotoxin limit applied, the standard curve and its acceptance, the positive product control recovery, and the result with its units. A number without the recovery figure cannot support a release decision.

**What is inhibition and enhancement testing?**

A product-specific validation showing your sample does not interfere with the endotoxin assay, done by spiking a known amount into the product and recovering it within the required window. It comes before routine testing rather than alongside it.

**Is a bioburden test the same as a sterility test?**

No. Bioburden counts viable organisms in a product before sterilisation and informs the sterilisation process. Sterility testing asks whether a sterilised product shows growth. They answer different questions at different points.

**Can a product be sterile and still fail an endotoxin detection kit?**

Yes, and that is the point of testing both. Sterilisation kills organisms and leaves their cell wall material, which is pyrogenic, so a sterile product can carry an endotoxin load well above its limit.

**What sets the endotoxin limit an endotoxin test kit reads against?**

The dose and the route of administration, through the standard formula in the compendial method. It is a property of the product's use rather than of the product itself, which is why the request has to state the intended dose.

## Culture microscopes

A culture microscope is used many times a day by everyone in the laboratory, which makes ergonomics and vessel fit matter more than optical specification. Two decisions settle most of it: the instrument must look up through the vessel base, and it must have phase contrast, because unstained living cells have almost no contrast of their own.

1. **Buy inverted, because the cells are on the bottom**: Cells grow on the vessel base with medium above, so the optics must look up through the plastic. An upright stand can be made to work with a long working distance condenser and is the wrong instrument for this job.
2. **Insist on phase contrast with matched annuli**: Unstained living cells are nearly invisible in brightfield. Phase contrast needs a ring in the condenser matched to each phase objective, and an instrument sold with phase objectives and a plain condenser produces a dim, haloed image nothing can fix.
3. **Check the condenser clears your tallest vessel**: A large flask that does not fit under the condenser cannot be imaged with phase contrast at all. Measure your tallest routine vessel and check the working distance before ordering, because this is discovered after delivery more often than it should be.
4. **Choose magnifications for the job people actually do**: Most culture work happens at low magnification: checking confluence, morphology and contamination. A four times and a ten times phase objective cover the great majority, with a twenty times for detail. Higher magnifications on a culture stand are rarely used.
5. **Decide whether a camera and screen earn their place**: A camera turns a subjective judgement into a shared, dated image and makes teaching and troubleshooting far easier. A screen instead of eyepieces suits a shared instrument and a cabinet, and it costs optical quality per unit of budget.

**Upright or an inverted microscope for cell culture, and where inverted microscopes differ?**

Inverted, always. Cells adhere to the vessel base with medium above, so the objective has to look up through the plastic. An upright stand cannot reach them in a normal vessel.

**Do I need phase contrast?**

Yes. Unstained living cells have almost no absorption contrast and are effectively invisible in brightfield. Phase contrast is what makes routine culture observation possible and it must be specified with matched condenser annuli.

**What working distance do I need?**

Enough for your tallest routine vessel under the condenser. Large flasks defeat short working distance condensers, and the result is that phase contrast cannot be used on exactly the vessels you use most.

**Is a camera worth it?**

For a shared instrument, usually. A dated image of a culture settles arguments about confluence and contamination, makes handovers possible and turns a judgement into a record that can be compared next week.

## Plasmid sequencing

Sequencing a plasmid used to mean designing primers and walking across the insert, which verified what you looked at and nothing else. Whole plasmid sequencing now costs little more and verifies the entire molecule, including the backbone nobody checks and the rearrangements that explain a construct behaving oddly. This page covers choosing a service and what to send.

1. **Sequence the whole plasmid, not just the insert**: Backbone mutations, lost promoter elements and rearrangements from propagation are real and invisible to a primer walk across the insert. Whole plasmid sequencing catches them for a marginal cost, and it is the reason the method has displaced walking.
2. **Use long reads where the construct has repeats**: Repeated elements, inverted terminal repeats and long homologous stretches cannot be resolved by short reads alone, and an assembly will quietly collapse them. For viral vector backbones and anything with repeats, long reads are the method rather than a refinement.
3. **Send the amount and purity they ask for**: Providers state a mass and a purity, and a preparation with residual RNA, protein or salt gives a failed run rather than a poor one. Quantify properly rather than relying on an absorbance reading alone.
4. **Send your map, and ask for the comparison**: The useful deliverable is an assembly compared against your expected sequence with the differences called, not a raw file. A service returning reads without an annotated comparison has handed you the work rather than done it.
5. **Verify before anything expensive depends on it**: Sequence the construct before a virus is made, before a stable line is generated and before a study starts. The cost of sequencing is trivial against the cost of discovering a construct was wrong three months in.

**Is whole plasmid sequencing worth it over primer walking?**

Almost always. It verifies the backbone and catches rearrangements that a walk across the insert cannot see, for a marginal additional cost, which is why it has largely replaced walking.

**When do I need long reads?**

When the construct contains repeats, inverted repeats or long homologous regions, which includes most viral vector backbones. Short read assemblies collapse repeats silently, which is the failure mode you are trying to avoid.

**What should the deliverable include?**

An assembled sequence, an annotated comparison against your expected map with every difference called, the coverage, and the raw reads. A file of reads with no comparison is not a verification.

**How much plasmid should I send?**

What the provider specifies, at their stated purity, quantified by a method that measures the DNA rather than everything that absorbs. Residual RNA is the commonest reason a sample reads as sufficient and then fails.

## Label-free microscopy

Two techniques appear in the same searches and answer entirely unrelated questions. One images chemistry in living tissue without a label by exciting molecular vibrations; the other maps electrical carrier distribution across a semiconductor surface with a probe. Both are specialist, both need an experienced operator, and confusing them wastes a purchasing conversation. This page separates them.

1. **Coherent Raman imaging maps chemistry without a stain**: It excites a molecular vibration and images the signal, so lipids, proteins and specific bonds can be mapped in living tissue with no label at all. For lipid biology and for tissue where staining is impossible or would perturb the sample, nothing else does this.
2. **Accept its chemical specificity limits**: The contrast comes from a vibrational band, so molecules sharing that band are not distinguished. It separates lipid from protein readily and two similar lipids far less readily, which is the trade against a labeled method's exact specificity.
3. **Scanning capacitance microscopy answers a semiconductor question**: It is a scanning probe method that maps carrier concentration and dopant distribution across a surface. It belongs in a materials or device laboratory, and it appears alongside biological microscopy only because both are called microscopy.
4. **Budget the operator, not only the instrument**: Both techniques need somebody who runs them regularly. A coherent Raman system without a laser-experienced operator, or a scanning probe without a probe-experienced one, becomes an expensive instrument nobody books.
5. **Use a core facility before buying either**: These are the clearest cases for shared access: expensive, intermittently used, and much better with an expert beside them. Buy only when the work is continuous and somebody's role includes running it.

**What does coherent Raman microscopy show?**

Chemical contrast without any label, by imaging molecular vibrations. It maps lipids and other bond-specific signals in living tissue, which is exactly what fluorescence cannot do without introducing a probe.

**How specific is label-free chemical imaging?**

Specific to a vibrational band rather than to a molecule, so it separates broad classes such as lipid from protein well and closely related species poorly. That is the trade against the exact specificity of a labeled method.

**Is scanning capacitance microscopy related?**

Only by the word microscopy. It is a scanning probe technique mapping carrier and dopant distribution in semiconductors, and it belongs in a materials laboratory rather than a biological one.

**Should I buy or use a facility?**

Use a facility unless the work is continuous. Both techniques depend heavily on operator experience, and an instrument without a regular operator is an expensive surface.

## Blotting apparatus

Blotting equipment is bought once and lived with for a decade, and the items that decide whether blots are even are the cheapest ones. A power supply rarely limits a blot; worn pads, an ill-fitting cassette and a tray that wastes antibody do. This page covers what to buy and what to replace on a schedule.

1. **Choose wet or semi-dry from the proteins you run**: A wet tank is slower, gentler and the most reliable for large proteins and for long transfers in a cold room. Semi-dry is much faster and can dry out on long runs. Laboratories running a wide size range usually keep a tank and use it for the difficult cases.
2. **Treat cassettes and pads as consumables**: Foam pads compress permanently and a cassette that no longer applies even pressure gives patchy transfer that looks like a gel problem. Date them when first used and replace them on that basis rather than waiting for a bad blot.
3. **Buy trays sized to the membrane**: Antibody incubation volume is set by the tray, and an oversized tray wastes expensive antibody on every blot. A tray that just fits the membrane, or a sealed bag, cuts antibody use substantially over a year.
4. **Get the power supply right once and forget it**: Constant voltage for transfer and constant current or voltage for the gel, with enough headroom for the tank's resistance. This is the item people compare and the one that least often limits a result.
5. **Plan for cold and for the bench around it**: Wet transfer generates heat and runs better cold, so a cold room, an ice pack insert or a cooled unit is part of the purchase. Add a rocker for incubations and a way to keep a membrane wet, because a dried membrane is a lost blot.

**Wet tank or semi-dry transfer?**

A wet western blot tank for large proteins, long transfers and the most reliable result; semi-dry for speed on routine sizes. Many laboratories own both and use the tank for the difficult cases.

**Why are my transfers patchy?**

Usually the pads or the cassette rather than the settings. Compressed pads apply uneven pressure, and paper cut larger than the gel lets current bypass the sandwich. Both are cheap and both are ignored for years.

**Does the western blot tray matter?**

For antibody cost, considerably. Incubation volume is set by the tray, so one sized to the membrane, or a sealed bag, cuts primary antibody use on every blot with no downside.

**What else limits a blotting bench?**

A rocker, enough trays to run several blots at once, a way to keep membranes wet, and cold transfer capacity. None is expensive and each of them is what actually slows a busy bench down.

**Does a western blot rocker matter?**

For the incubations, yes, more than people expect. An antibody solution that is not moving depletes locally, and the result is a blot that is stronger where the liquid happened to circulate. A slow orbital or rocking platform that keeps the whole membrane covered is enough; speed is not the point, coverage is.

**Is a nitrocellulose membrane western blot better for small proteins?**

Often, because the smaller pore size holds a low molecular weight target that can pass through a larger pore membrane during a long transfer. The other half of the answer is the transfer time: a small protein is lost by transferring for as long as a large one needs, whichever membrane is used.

## CMO against CDMO

The distinction is simple and is routinely got wrong because programs overestimate how finished their process is. A contract manufacturer makes what you give it to a process you have already defined; a development and manufacturing organization develops that process first. Choosing the first with a process that is not ready produces a delay that looks like a partner problem and is not.

1. **Be honest about how defined your process is**: A finished process has fixed unit operations, defined parameters and ranges, qualified analytical methods and a specification. If any of those is still moving, you have a development project, and giving it to a manufacturer as if it were finished makes the development happen anyway, slower and at a higher rate.
2. **Check whether your analytics transfer as well as your process**: Methods have to be qualified at the receiving site, and programs routinely plan the process transfer and discover the analytical transfer afterwards. That is where the months go, and it is a development activity whoever performs it.
3. **Match the partner to your scale and your format**: A site running your format weekly will spot a problem sooner than one running it for the first time at a better rate. Ask what they run routinely rather than what they can run, and ask for a reference program of your size.
4. **Read the slot as seriously as the price**: Capacity in eighteen months is not capacity, and a slot without penalty terms is an intention. Confirm the slot, the campaign length and what happens if your material is late, because those terms describe how real it is.
5. **Settle ownership of process knowledge in the contract**: Who owns process improvements, whether you can take the process elsewhere and what documentation transfers at the end are terms, not defaults. A development partner that owns your process is a partner you cannot leave.

**What is the difference between CMOs and CDMOs?**

A contract manufacturer makes a process you have defined and transferred. A development and manufacturing organization develops or finishes the process first and then makes it. The extra capability is development, and it is charged for.

**How do I know my process is ready for a CMO?**

Unit operations fixed, parameters and ranges defined, analytical methods qualified and a specification agreed. If any of those is open, the development will happen wherever you go, and a manufacturer is not set up to do it well.

**Is a CDMO more expensive?**

For the same batch, generally yes, because development capability is being paid for. For a program whose process is not finished, it is usually cheaper overall than paying a manufacturer to discover the process during a campaign.

**What is the most common mistake?**

Underestimating how much analytical transfer and how many engineering runs are needed, and planning straight into a clinical batch. That plan produces an investigation rather than material, at either kind of partner.

## Cell culture equipment

A cell culture laboratory can be equipped in the wrong order, and the usual mistake is buying the instrument that does the interesting experiment before the containment and the incubator that make any experiment possible. Both have lead times and installation requirements, and nothing else matters until they are in and certified. This page sets out the order and the consumables that go with it.

1. **Containment and incubation first**: A biosafety cabinet certified in place, and a carbon dioxide incubator with a decontamination cycle. Both have lead times, both need siting decisions about heat and airflow, and neither can be improvised. Nothing else in the room matters until they are working.
2. **Then the separations used every day**: A benchtop centrifuge with a swing-out rotor for tubes and plates, refrigerated if primary cells or proteins are involved. This is the item a growing laboratory queues for first, so size it for the people rather than for today's samples.
3. **Then cold storage with alarms**: A laboratory refrigerator, a minus twenty and a minus eighty with monitoring and an alarm somebody receives, plus nitrogen storage for cell banks. Cell stocks are the laboratory's real asset and the failure plan belongs in place before the first vial goes in.
4. **Choose vessels and plates from the assay, not the catalog**: Surface treatment, well geometry, optical bottom and lid design all change results. An assay read on an imager needs an optical bottom; a suspension culture needs an untreated surface. Standardize on as few formats as the work allows.
5. **Buy the small things that stop work**: A water bath or bead bath, a pipette controller per cabinet, aspiration with a trap, a cell counter, and enough plasticware to work for a term. A laboratory with an incubator and no serological pipettes cannot start.

**What comes first, the cell culture hood or the incubator?**

A certified biosafety cabinet and a carbon dioxide incubator. Both have lead times and siting requirements, and no culture work can begin until they are installed and the cabinet has been certified in place.

**How do I choose cell culture plates?**

From the assay. Surface treatment decides attachment, optical bottoms are required for imaging, and well geometry changes edge effects and evaporation. Standardize on as few formats as the work allows to keep the stock manageable.

**How much cold storage does a new culture laboratory need?**

More than current need, with monitoring, alarms and a written failure plan before the first vial. Cell banks are the irreplaceable asset, and the day a freezer fails is the wrong day to find out where their contents would go.

**What is most often forgotten?**

The consumables and the small equipment: pipette controllers, aspiration with a trap, a counter, and enough plasticware. They are cheap, they have lead times, and their absence stops a fully equipped laboratory from working.

## Mouse and IgG ELISA

Species-specific kits are the most common ordering error in immunoassay, because the target name is the same and the reagents are not. A mouse cytokine kit will not measure the human protein and vice versa, and an immunoglobulin kit is specific not only to species but often to subclass. This page covers choosing correctly and qualifying whatever you choose in your own matrix.

1. **Match the species exactly, every time**: Cytokines are less conserved than their shared names suggest, and a kit raised against the mouse protein generally will not detect the human one. Check the species on the kit against the species of your sample before anything else, because this is the error that wastes a whole plate.
2. **For immunoglobulin kits, check the subclass and the specificity**: A total immunoglobulin kit and a subclass-specific one answer different questions, and an anti-species reagent may or may not distinguish subclasses. For antibody quantitation in a culture supernatant the subclass usually matters.
3. **Match the range to the expected concentration**: Immunoglobulin in a supernatant and a cytokine in serum sit orders of magnitude apart. Pick the kit whose standard curve brackets your expected values near its middle, where precision is best, rather than diluting to fit a convenient kit.
4. **Qualify in your own matrix**: Spike recovery and dilution linearity in the actual sample type. A kit validated in serum used on culture medium, lysate or a tissue extract is unvalidated until you demonstrate the matrix is not interfering.
5. **Fix handling before the first plate**: Tube type, time to centrifugation, storage temperature and freeze-thaw all move immunoassay results, and several cytokines are released during clotting. Write the protocol down and apply it to every sample in the study.

**Can a mouse ELISA kit measure the human protein?**

Generally no. These proteins are less conserved than their shared names suggest, and the antibody pair is raised against one species. Check the species on the kit against your sample before ordering.

**Total IgG or subclass-specific?**

It depends on the question. For quantifying a monoclonal in a supernatant, the subclass-specific kit is usually the right one; for total immunoglobulin in a fluid, the total kit. They are not interchangeable and the standards differ.

**Why does my kit saturate?**

The range does not match the sample. Immunoglobulin concentrations sit far above cytokine ones, so a kit chosen by target name rather than by expected concentration will either saturate or read below its floor.

**Do I have to qualify a commercial kit?**

In your own matrix, yes. Spike recovery and dilution linearity take one plate and tell you whether the matrix suppresses or enhances the signal, which no datasheet can tell you about your samples.

## Equipment financing

Laboratory instruments are bought with a capital budget that arrives once and consumed by running costs that arrive every year, which is why financing is worth thinking about rather than defaulting to. The comparison only works if service, consumables and the end of term are in it, and those are exactly what a headline monthly figure leaves out.

1. **Understand what each route does to ownership**: An outright purchase gives you the asset and the residual value. A finance lease is a purchase spread over time and usually ends in ownership. An operating lease is a rental where the provider keeps the asset and takes the residual risk. Those are three different things and only the third protects you from obsolescence.
2. **Put service and consumables in the comparison**: On most instruments the five year service and consumable bill exceeds the purchase price, and proprietary consumables are where a cheap instrument becomes expensive. Compare total cost over a realistic life, not the acquisition figure.
3. **Read what happens at the end of the term**: Return condition requirements, refurbishment charges, automatic renewal clauses and the buyout figure are where lease costs hide. A return condition you cannot meet turns a clean exit into a bill.
4. **Check who carries the risk of failure and obsolescence**: A lease that continues whatever happens to the instrument leaves you paying for a broken asset. Ask specifically what happens if the instrument fails, is superseded, or the vendor stops supporting it.
5. **Decide from the instrument's likely life, not the cash flow**: Instruments that will be used for a decade suit purchase; those in a fast-moving area, or needed for a fixed-term project, suit an operating lease. Letting cash flow alone decide is how a laboratory ends up renting something it should have bought.

**Lease or buy laboratory equipment?**

Buy what you will use for a decade and can service; lease what will be superseded or is needed for a fixed term. The decision follows from the instrument's useful life rather than from the monthly figure.

**What does a headline monthly figure leave out?**

Service, consumables, insurance, the end of term condition and any buyout. On most instruments the running costs exceed the capital cost over five years, so a comparison without them is not a comparison.

**What clauses matter most in a lease?**

Return condition and refurbishment charges, automatic renewal, the buyout figure, and what happens if the instrument fails or is discontinued. Those four decide whether a lease ends cleanly.

**Is refurbished equipment worth considering?**

For mechanically simple items with a warranty and a service history, frequently yes. For anything with optics, detectors or a regulated role, the service history and parts availability matter more than the saving.

## Oligonucleotides

An oligonucleotide is priced by the base, quoted by the synthesis scale and delivered at a purity someone chose on your behalf unless you state one. Those three variables move the cost of the same sequence by an order of magnitude, and they are the difference between a primer that works in a routine reaction and a probe that has to survive a clinical assay. This page sets out what actually changes when the order form changes, so a specification can be written once and reused.

1. **Set the scale from the use, not from the catalog default**: Synthesis scale decides how much material you receive, and most laboratories order far more than a PCR program will ever consume. A small scale covers years of routine primer use; a probe going into a validated assay, or a guide going into a screen, is where a larger scale and a second synthesis lot earn their cost. Ordering the smallest scale that covers the work removes freezer clutter and makes reordering a deliberate act.
2. **Choose purification against the reaction, not against a grade name**: Desalted material is right for standard amplification. Cartridge and HPLC purification remove the truncated failure sequences that matter when a labeled probe, a long assembly fragment or an antisense construct has to behave predictably. Gel purification is the choice when full length is the only acceptable product. Every step up in purification costs yield, so buy it where a shortened sequence would change a result.
3. **Declare modifications early and in full**: A five prime label, a three prime quencher, a phosphorothioate backbone, a spacer or a locked base each change the coupling chemistry and often the minimum scale. Suppliers quote them differently, and a modification that looks minor can double a lead time. List every modification with its position in the sequence and ask which ones force a purification you would not otherwise buy.
4. **Order in plates when the set is a set**: A screening panel, a tiling set or a library of variants belongs in a plate at a stated concentration in a stated buffer, not in ninety six individual tubes. Plate orders arrive normalised, cost less to handle and make the downstream automation trivial. State the buffer explicitly: a low EDTA tris buffer is the common default and it is not the same as water for long term storage.
5. **Fix a storage and resuspension rule for the laboratory**: Dried oligonucleotides are stable for a long time at low temperature; resuspended stocks are not, and repeated freezing and thawing is the usual cause of a primer that stopped working. Agree one resuspension concentration, one buffer and one aliquoting habit, and write them where the plate lives.

**What actually drives the price of an oligonucleotide?**

Scale, purification and modification, in that order, with length as a secondary factor. Two quotations for the same sequence are rarely comparable until all three are stated, which is why a written specification is worth more than a price list.

**When is desalted material not good enough?**

When a truncated sequence would change the result: labeled probes, long assembly fragments, antisense and therapeutic constructs, and anything going into a validated method. Routine amplification primers are usually fine desalted.

**Should I order oligo pools or individual tubes?**

A pool is the right product when the set is used together and never individually, for example a screening library or a tiling panel. Individual synthesis is right when each sequence has to be tracked, quantified and reordered on its own.

**What should custom DNA oligos be resuspended in?**

A buffered low EDTA tris solution rather than water for stocks you intend to keep, with working dilutions made fresh. Fix one concentration and one buffer across the laboratory so nobody has to interpret a tube label.

**Does the scale on the order tell me how much I get?**

No. Scale is the amount of the first base loaded; delivered yield falls with length. Read nanomoles delivered on the specification sheet and calculate the working concentration from that.

**When do I need more than desalting?**

Where every molecule matters: cloning with long overhangs, duplexes that must anneal stoichiometrically, modified oligonucleotides where the modification is the point, and anything used as a therapeutic model. Amplification primers rarely need it.

## Liquid handling robots

Most laboratories choose a liquid handler on channel count and speed, and most of the disappointment that follows is about deck real estate, tip logistics and what happens when a run fails halfway through. A machine that pipettes faster than anyone needs is easy to buy; a machine that runs your assay unattended, recovers from an error without ruining a plate and produces a record someone can audit is a specification problem. This page is about that specification.

1. **Draw the deck before you compare instruments**: Lay out every labware position the method needs at once: source plates, destination plates, reagent troughs, tip racks, a waste chute and any on deck device such as a shaker, a magnet or a heater. Count the positions. Most methods fail the deck test long before they fail the speed test, and a platform that needs the operator to reload mid run is not walk away automation.
2. **Decide air displacement against positive displacement**: Air displacement covers the majority of aqueous work and uses cheap disposable tips. Positive displacement handles viscous, volatile and dense liquids that defeat an air cushion. Mixed workloads usually justify a platform that accepts both, or an honest decision that the difficult liquids stay manual.
3. **Cost the tips, not just the instrument**: Filter tips, conductive tips for liquid level sensing and low retention surfaces each carry a per run cost that quickly exceeds the depreciation on the machine. Ask what a year of the intended run rate costs in consumables, and whether the platform accepts a second source of tips or locks you to one.
4. **Test error recovery with a deliberate failure**: During the demonstration, pull a tip rack, block a channel or remove a plate and watch what the software does. The question is whether the run pauses recoverably with the plate state recorded, or aborts and leaves you unable to say which wells were dispensed. That behavior, not throughput, decides whether the machine can run overnight.
5. **Check the data path and the audit trail**: Confirm how the run log leaves the instrument, whether it records per well what was aspirated and dispensed, and whether user actions are attributable. If the results will ever support a regulated decision, electronic record and signature expectations apply to the instrument software as much as to the laboratory system it feeds.
6. **Separate scheduling from method writing**: Ask who will write methods after the vendor's engineer leaves. A platform that only the supplier can reprogram becomes a fixed function machine within a year. Look for a method editor a scientist can learn, and budget the training rather than hoping.

**How many channels does a liquid handling machine actually need?**

Fewer than most buyers assume. Eight channels plus a plate handling head covers a very wide range of plate work; ninety six and three hundred and eighty four channel heads pay off only when the same transfer is repeated across many plates. Deck capacity limits throughput long before channel count does.

**Is an automated colony picker the same class of machine as a pipetting robot?**

No. Colony picking and single cell dispensing are imaging and selection problems with a pipetting step attached, and the instruments are specified on optics, picking accuracy and viability rather than on transfer volume. Buying a liquid handler and expecting it to pick colonies is a common and expensive mistake.

**What does a liquid handling robotics quotation usually leave out?**

Tips and labware for the intended run rate, method development time, a service contract, the qualification visit and any on deck devices the method needs. The instrument line is often the smaller half of the first year cost.

**Should the platform be validated?**

If its output supports a regulated decision, yes, and the requirement should be written into the purchase. Installation, operational and performance checks are cheaper to specify before delivery than to retrofit after the first audit finding.

## Sample and asset tracking

Three different products get bought under the same request. One tracks vials and their lineage. One tracks instruments, their calibration and their service history. One schedules and bills access to shared equipment. They overlap enough to be confused in a demonstration and differ enough that the wrong choice is discovered about six months in, when someone asks a question the system was never built to answer. This page separates them.

1. **Write down the question you cannot answer today**: Is it where a particular vial is and what it came from? Is it when an instrument was last calibrated and who serviced it? Is it who used the cytometer last month and what they should be charged? Each question maps to a different product, and a laboratory that cannot state which one it has is not ready to buy.
2. **Follow a sample end to end on paper first**: Take one real sample and write every point where it is split, moved, consumed or destroyed, and who records that. The points where nobody records anything are the requirements. A system that does not capture those points will be filled in retrospectively, which is worse than paper.
3. **Decide whether lineage or location is the point**: Location tracking says which freezer, rack and position holds a tube. Lineage says which parent it was derived from and what was done to it. Most laboratories need both, but only lineage survives a question asked two years later, and many cheap trackers only do location.
4. **Test the search that matters, not the dashboard**: Ask the vendor to find every aliquot descended from one collection, or every instrument due for calibration in a named month, in front of you. Dashboards demonstrate well and answer nothing. The retrieval query is the product.
5. **Settle the record requirements before the license**: If any output feeds a regulated decision, the system needs attributable, time stamped, unalterable records and a defined user access model. That is a different tier of product from a shared spreadsheet with a web front end, and retrofitting it is not realistic.
6. **Plan the exit before the entry**: Ask how data leaves: full export, in an open format, including the audit trail. A system whose export is a rendered report rather than structured data has taken the laboratory's records hostage, and that becomes apparent only when you try to move.

**Is sample tracking just a small LIMS?**

No. A laboratory information management system is organized around tests, workflows and results; a sample tracker is organized around the physical item and its lineage. Laboratories that buy the wrong one find themselves fighting a data model rather than using it.

**What does core facility management software do that the others do not?**

It schedules shared instruments, records who used what for how long, and turns that into a bill or a cost allocation. Tracking vials is not its job, and asking it to do so produces a clumsy result.

**Does lab asset management software need to hold calibration records?**

If the instrument produces data anyone relies on, yes. An asset register that lists serial numbers but not calibration status, service history and out of tolerance events answers the finance question and none of the scientific ones.

**Is free software ever the right answer?**

For a small group tracking location only, sometimes. It stops being the right answer the moment a record has to be attributable and unalterable, because that is precisely what the free tier usually omits.

## Cell line development

A cell line development project produces two things: a clone that expresses what you need, and a body of evidence that the clone is what you say it is and will still behave that way at the end of a production run. Laboratories that buy only the first find out during a technology transfer that they bought half a project. This page is about specifying both, and about the differences between mammalian, microbial and gene edited routes that the word custom hides.

1. **Choose the host from the molecule, not from habit**: Mammalian hosts give human like glycosylation and are the default for complex secreted proteins and antibodies. Microbial hosts give speed and cost for proteins that fold without those modifications. The wrong host is recoverable only by starting again, so the decision belongs before any construct is built.
2. **Fix the selection and screening funnel in advance**: Agree how many clones enter screening, on what criteria they are ranked, and how many go forward. Titre alone is a poor ranking: growth profile, product quality and behavior at the intended scale matter as much, and a clone that wins on titre in a plate frequently loses in a bioreactor.
3. **Demonstrate clonality in a way someone can check**: A single cell origin has to be evidenced, not asserted. Imaging at deposition, a documented limiting dilution with statistics, or both, is what the record needs. This is the item most often missing when a line arrives from a partner and the most expensive to reconstruct.
4. **Build the bank structure before you need it**: One master bank, working banks drawn from it, and physical separation between storage locations. Decide the vial counts against the projected campaign schedule, and decide them once, because a bank that runs out forces a new bank and a new comparability exercise.
5. **Test the bank to a written panel**: Identity, viability, recovery, sterility, mycoplasma and adventitious agent testing as the product type demands, with the panel agreed before the vials are filled. Testing after the fact against a panel nobody wrote is how a bank becomes unusable.
6. **Run stability to the limit of the intended process**: Culture the line past the generation count any campaign would reach and show that expression and product quality hold. A line stable through ten passages and failing at forty is a line you will discover at the worst possible moment.

**How long does a mammalian cell line project take?**

Realistically several months from construct to a banked, characterised line, with screening and stability consuming most of it. Compressed timelines usually compress the stability work, which is the part that protects you later.

**What makes a cell bank a GMP cell bank?**

Manufacture under a quality system, with controlled materials, documented procedures, defined release testing and retained records, rather than any property of the cells. The same clone can be banked at research grade and at GMP grade, and the difference is entirely in the documentation and the environment.

**Why is clonality evidence so important?**

Because a heterogeneous population drifts, and a drifting population invalidates everything measured against it. Regulators and partners ask how single cell origin was demonstrated, and an answer of limiting dilution alone is increasingly treated as insufficient without supporting evidence.

**Does an edited line need different banking?**

It needs the same banking plus evidence about the edit: on target confirmation, characterisation of the insertion or disruption, and an assessment of off target effects proportionate to the use. The banking structure itself does not change.

## IVT mRNA synthesis

In vitro transcription looks like a one tube reaction and behaves like a small manufacturing process. The template decides the ends of the product, the polymerase and the nucleotide mix decide the impurity profile, and the purification decides whether what you deliver stimulates the cells you wanted or the innate immune sensors you did not. This page walks the decisions in the order they actually constrain each other.

1. **Design the template with the product's ends in mind**: The transcript begins where the promoter says and ends where the template ends, so linearisation site and completeness of digestion define the three prime end. Incomplete linearisation gives long runoff products that are difficult to remove later. Verify the cut before every transcription rather than trusting a stock preparation.
2. **Choose the capping strategy deliberately**: Co transcriptional capping with a cap analogue is simpler and gives a defined structure in one step; enzymatic capping after transcription can give higher capping efficiency at the cost of another unit operation. Both are defensible. What is not defensible is being unable to state your capping efficiency.
3. **Decide how the tail is added**: Encoding the tail in the template gives a defined, reproducible length; enzymatic tailing gives a distribution. Defined length is easier to control and easier to characterise, and it removes an assay you would otherwise have to run on every lot.
4. **Control the double stranded byproduct**: Transcription generates double stranded species that are potent innate immune activators, and they are the usual explanation for an mRNA that works in a cell free assay and fails in cells. Modified nucleotides reduce the problem; purification that specifically removes double stranded species addresses it directly.
5. **Purify against the intended use**: Precipitation is adequate for a transfection control. Chromatographic purification is what a potency sensitive application needs, and it is the step that usually separates a research preparation from one that behaves consistently in animals or in people.
6. **Write the release panel before the first lot**: Identity, integrity, concentration, capping efficiency, tail length, residual template, residual protein and double stranded content, with acceptance criteria agreed in advance. A panel written after a failure is a panel written to explain it.

**Why does my mRNA work in vitro and not in cells?**

The usual cause is double stranded byproduct triggering innate immune sensing, followed by incomplete capping. Both are process problems rather than sequence problems, and both are fixed by purification and capping strategy rather than by redesigning the construct.

**Is a kit sufficient for preclinical work?**

For process controls and early cell work, often yes. Once potency matters or an animal study depends on the material, the purification and characterisation a kit does not include become the difference between interpretable and uninterpretable results.

**Should the tail be encoded or added enzymatically?**

Encoded, where the construct allows it. A defined length is reproducible, characterisable and removes a source of lot to lot variation that is otherwise difficult to control.

**What changes at manufacturing scale?**

Everything about control rather than chemistry. Raw material qualification, a validated linearisation, in process controls, a defined purification train, environmental control and a documented release panel. The reaction is the smallest part of the transfer.

## Plasmid construction

There is no best cloning method, only a method that suits the number of parts, the sequence constraints and whether the construct is a one off or a template for a series. Choosing badly costs weeks of screening colonies that were never going to be right. This page maps the common routes onto the jobs they are good at, and says plainly where buying the construct beats building it.

1. **Count the parts and the reuse**: One insert into one vector is a job for any method. Four or more parts, or a series of constructs sharing a part library, is where a standardized assembly with defined junctions repays the setup. If you will build this family again, build the library once.
2. **Check the sequence for the method's blind spots**: Type two restriction assembly needs the chosen site absent from every part, and a single internal occurrence silently ruins the design. Overlap methods dislike repeats and strong secondary structure. Screen the sequence against the method before ordering anything.
3. **Design mutagenesis around template removal**: Amplification based mutagenesis produces a mixture of new product and methylated parental template, and the digestion that removes the parental plasmid is the whole game. Skipping or under running it produces colonies that sequence as the original, which is the most common wasted week in this area.
4. **Match the ligase to the substrate**: Standard double stranded ligation, single stranded ligation and adaptor joining are different enzymes with different buffers and different optimal temperatures. Using a general purpose enzyme on a specialist substrate produces a low yield that is usually blamed on the insert.
5. **Sequence the whole construct, not the junctions**: Junction sequencing was a reasonable economy when reads were expensive. Whole plasmid sequencing now costs little and catches backbone rearrangements, unintended point changes and the wrong parent entirely, all of which junction reads miss.
6. **Decide build against buy honestly**: A synthesised, sequence verified construct delivered ready to use frequently costs less than the labor of building it, particularly for difficult sequences or when a catalog clone already exists. Price the bench time before defending the build.

**Which assembly method should a laboratory standardise on?**

Whichever matches its most common job. Groups building multipart constructs repeatedly benefit from a standardized type two site assembly with a maintained part library; groups making occasional single insert constructs rarely need more than overlap assembly or ordering the construct.

**Why do my site directed mutagenesis reactions give the original sequence?**

Almost always incomplete removal of the methylated parental template. Check the enzyme is active, extend the digestion and confirm that the input plasmid came from a strain that methylates, because template from a methylation deficient strain will not be cut at all.

**Is whole plasmid sequencing worth it every time?**

Yes, for anything that will be used more than once or shared. The cost is small against the cost of discovering a backbone rearrangement after six months of experiments built on the construct.

**When does an off the shelf ORF clone beat a custom build?**

Whenever the sequence, the tag position and the vector backbone you need already exist in a catalog. The saving is not only the build but the sequence verification and the risk, and the main thing to check is the exact isoform and the license attached to it.

**Should I clone it or have it synthesised?**

Synthesis has become cheap enough that building a routine construct by hand is often more expensive once time is counted. Clone by hand where you already have the parts and the skill; buy a sequence-verified construct where turnaround matters.

**How much of a plasmid should I sequence?**

All of it. Verifying only the insert leaves junctions, promoter and backbone unchecked, and whole-plasmid sequencing is now cheap enough that there is little reason to do less.

## Plasmid production

A plasmid preparation that is fine for transfecting a dish is not fine for making a viral vector, and neither is fine for dosing a person. The chemistry barely changes across those tiers; what changes is the host strain control, the documentation, the impurity limits and who is accountable when a lot fails. Buying the wrong tier is the most expensive mistake in this category because it is usually discovered downstream.

1. **Name the downstream use first**: Transfection in research cells, viral vector manufacture, or direct administration each set different limits on endotoxin, host residuals and documentation. State the use in the enquiry, because a supplier quoting without it will quote the cheapest grade and both parties will be surprised later.
2. **Specify the supercoiled fraction, not just the mass**: Yield in milligrams says nothing about how much of that mass is the intact supercoiled form that performs in transfection. Set a minimum supercoiled fraction with a stated method, and treat open circular and linear content as a specification rather than an observation.
3. **Set the endotoxin limit against the application**: Bacterial endotoxin is the impurity that most often makes a preparation unusable in cells and animals, and the acceptable level differs by orders of magnitude between a research transfection and a clinical intermediate. Agree the limit and the test method up front.
4. **Control the host and the bank**: A production lot should come from a characterised, banked host strain carrying a sequence verified construct, not from a colony someone streaked. Ask for the bank documentation and the construct sequence evidence with the quotation.
5. **Agree the release panel and the retain policy**: Identity by sequence or restriction pattern, concentration, supercoiled fraction, endotoxin, residual host protein, residual host nucleic acid, residual RNA and sterility as the grade requires. Agree who retains samples and for how long.
6. **Plan the transfer between grades early**: Moving from research to graded material changes the host, the process and sometimes the construct. Doing that transfer while a downstream program waits is where timelines are lost, so run the graded lot before it is on the critical path.

**What makes a plasmid lot GMP plasmid manufacturing?**

Manufacture under a quality system with qualified materials, controlled and documented procedures, defined in process controls, an agreed release panel and retained records. It is a statement about the process and its evidence, not about the purity number alone.

**Why does supercoiled fraction matter so much?**

Because the supercoiled form transfects and expresses far more efficiently than the relaxed or linear forms. Two lots with the same mass and very different supercoiled fractions behave differently in the same experiment, which is why the fraction belongs in the specification.

**Is endotoxin removal always necessary?**

Not for bacterial work, and often not for robust immortalised lines. It becomes essential for primary cells, immune cells, animal work and anything clinical, where endotoxin produces a biological response that is easily mistaken for the effect under study.

**How far ahead should a graded lot be ordered?**

Considerably further than a research preparation, because the bank, the documentation and the release testing all take time before the fermentation does. Treat lead time as part of the specification and confirm it in writing.

## Viral vector CDMOs

Vector manufacturing slots are scarce and expensive, and the usual failure is not capacity but mismatch: a partner whose platform, analytics and scale do not fit the program, discovered after a campaign has been paid for. The questions that separate partners are about process ownership, analytical methods and what happens when a batch misses specification. This page is those questions.

1. **Decide what you are buying: a platform or a pair of hands**: A platform partner runs their established process and gives you speed and predictability, at the cost of process ownership and sometimes of the ability to move. A contract manufacturer running your process gives you control and portability, at the cost of time and of doing the development yourself. Both are valid. Confusing them is not.
2. **Pin down every titre method by name**: Genome titre, capsid or particle titre and infectious or transducing titre measure different things and differ by large factors. Two partners quoting a titre without naming the method are not quoting the same number. Insist on the method, the standard used and the acceptance range.
3. **Ask about empty and partial capsids explicitly**: For adeno associated vectors the fraction of capsids carrying no genome or a partial one drives the dose and the safety profile, and separating them is a real unit operation with a real yield cost. A process that does not address it is quoting a number that will not survive the first regulatory conversation.
4. **Review the analytical package, not just the process**: Identity, potency, purity, residual host protein and nucleic acid, residual plasmid, replication competent vector testing and sterility. Ask which methods are established at the partner, which are transferred and which are subcontracted, because subcontracted methods dominate turnaround.
5. **Model scale honestly against the intended dose**: Work backwards from doses required, dose level and the vector needed per dose to the batch size that program implies, then ask whether the partner has run that scale on that platform. Scale up in vector manufacture is not linear and a partner's largest demonstrated batch matters more than their installed volume.
6. **Negotiate failure before success**: What happens if a batch fails release, who pays, how quickly a reslot is available, and what data you receive from a failed run. These terms are far easier to agree before a campaign than during one.

**Should viral vector production use a suspension or an adherent process?**

Suspension scales predictably and is the default for anything heading past early clinical work. Adherent processes remain common at small scale and for legacy comparability, but moving from adherent to suspension later is a process change with comparability consequences, so decide early.

**Which titre number should a contract be written against?**

The one the dose is calculated from, stated with its method. For gene therapy that is usually a genome titre with a defined standard, accompanied by an infectivity ratio. A contract against an unnamed titre is a contract against nothing.

**How early should analytical development start?**

Before manufacturing, not alongside it. Methods that are not ready when a batch is released force retrospective testing on retained samples, which delays everything and occasionally cannot be done at all.

**Can a process be moved between partners?**

Only if it was written down in a form that transfers, and only if you own it. Platform processes are usually not portable, which is a reasonable trade to make knowingly and a painful one to discover.

**What should I ask a vector provider for?**

Step yields on a comparable construct rather than a final titre, which step they consider the bottleneck, and the qualified analytical package: titre method, empty to full ratio, replication competence and impurities.

## Blot and gel imagers

A blot imager is bought to make an image and used to make a number, and the gap between those two jobs explains most of the disappointment. Saturation is invisible on screen, chemiluminescent signal decays while you decide on an exposure, and film habits carried into a digital instrument produce data that cannot be defended. This page sets out what to specify if the output will ever be quantified.

1. **Decide whether you are quantifying or documenting**: Documenting a presence or absence tolerates almost any imager. Quantifying a change requires linearity across the range of your samples, a detector that reports saturation, and ideally a normalisation strategy that does not rely on a single housekeeping band. Buy for the harder of the two jobs you actually do.
2. **Ask for the linear range, in writing**: Every camera has a range over which signal is proportional to input, and it is narrower than the specification headline suggests. Ask the vendor to demonstrate a dilution series on your own samples and to show where proportionality fails. That experiment separates instruments more cleanly than any brochure figure.
3. **Choose chemiluminescence or fluorescence deliberately**: Chemiluminescence is sensitive and familiar but the signal changes with time, which makes multiplexing and precise quantitation awkward. Fluorescent detection is stable, genuinely multiplexed and easier to quantify, at the cost of different secondary reagents and usually a more expensive instrument.
4. **Check saturation reporting and exposure control**: An instrument that flags saturated pixels and offers automatic or signal accumulating exposure prevents the single most common quantitation error. One that simply brightens the display does not. This feature is worth more than a nominal sensitivity advantage.
5. **Confirm gel modes if the same box must do both**: Nucleic acid gel imaging needs the right excitation and emission filters for the stain in use, and safer stains often need different optics from older intercalating dyes. A blot imager without the correct filter set will image gels badly and expensively.
6. **Check how images and analysis leave the instrument**: Raw, unscaled image files must be exportable, and the analysis settings must be recorded with the result. An instrument that exports only an adjusted picture makes it impossible to show later that the number was not produced by the display settings.

**Is a cooled camera necessary?**

For chemiluminescent quantitation and for long exposures, yes, because thermal noise is what limits faint band detection. For bright fluorescent gels and routine documentation, an uncooled sensor is often sufficient and considerably cheaper.

**Why do phosphor screens still appear in laboratories?**

Because their linear response spans a very wide range, which makes them the reference method where quantitation over several orders of magnitude matters. They are slower and require their own reader, which is why they persist alongside cameras rather than instead of them.

**Can one instrument cover a western blot gel and a blot?**

Frequently yes, provided the filter sets cover both the chemiluminescent or fluorescent blot workflow and the nucleic acid stain in use. Confirm the specific stain, because filter compatibility is where a combined purchase usually disappoints.

**Is a cell imager, or an in cell western blot reader, the same category?**

No. Cell imaging instruments are microscopes with automated stages and analysis, specified on objectives, illumination and throughput. They appear in the same budget conversation and solve a different problem, and conflating the two leads to buying neither well.

## Residual impurity assays

Residual impurity assays are bought as kits and used as release tests, and the gap between those two ideas is where regulatory findings come from. A generic host cell protein immunoassay measures what its antibodies happen to recognize, which may be a small fraction of what your particular process leaves behind. This page is about what makes such an assay valid, and what the neighbouring tests are actually for.

1. **Match the kit to the expression system, then stop trusting it**: A kit raised against one cell line's proteome is a starting point for a process using that line and meaningless for another. Even within the right line, the specific clone, media and process shape the impurity profile, so the kit is a candidate rather than an answer.
2. **Demonstrate antibody coverage against your own process**: Coverage analysis, typically by two dimensional separation and immunoblotting or by mass spectrometry comparison, shows what fraction of the impurities present the antibodies actually detect. Without it, a low reported value may mean a clean product or a blind assay, and nobody can tell which.
3. **Establish dilutional linearity and spike recovery in your matrix**: The product itself interferes. Run the assay across dilutions of real samples and confirm the result is consistent, and spike known material to confirm recovery. A dilution dependent answer is a warning that the matrix is defeating the assay.
4. **Treat residual nucleic acid as a separate test**: Residual host DNA is measured by amplification or hybridisation rather than by immunoassay, and it has its own limit and its own sample preparation. It is a different question from residual protein and needs its own method development.
5. **Add a leachable test where the process uses affinity capture**: Affinity ligands shed, and the shed ligand is an impurity in its own right with its own assay. If the purification train includes an affinity step, the corresponding leachable assay belongs in the panel.
6. **Decide when profiling replaces counting**: A plate assay gives one number. Mass spectrometry based profiling identifies which proteins persist, which is what you need when a single stubborn impurity drives the total or when the total will not come down.

**Can a generic HCP kit be used for release testing?**

Only with documented coverage against the specific process, and usually only in early development. Programs moving forward typically develop a process specific assay, because the generic reagent cannot be shown to see what their process leaves behind.

**What does coverage analysis actually involve?**

Separating the host cell proteins present in a null run of your process, then showing which of them the assay antibodies bind. It is a piece of analytical development rather than a kit insert calculation, and it is the item most often missing from an early package.

**Why measure residual DNA separately?**

Because it is a different impurity class with a different limit and a different detection chemistry. Protein immunoassays do not see it, and the sample preparation that recovers protein frequently destroys or fails to recover nucleic acid.

**Are catalogue cytokine kits relevant here?**

Only by coincidence of format. They use the same plate technology for a completely different purpose, measuring an analyte of biological interest rather than an impurity, and they are validated against different expectations.

## Assay development

An assay is a machine for turning a sample into a decision, and the development project should be scoped from the decision backwards. Teams that start from a pair of antibodies and a plate usually end with an assay that works in buffer, drifts in the real sample matrix, and cannot be transferred to anyone else. This page sets out what to specify and what a finished development should hand over.

1. **State the decision and the required precision**: What will change depending on the number, and how small a difference must be detectable for that decision to be made? Those two answers set the required range and precision, and they often reveal that a simpler or a more sensitive format is needed than was assumed.
2. **Define the matrix before the reagents**: Serum, plasma, lysate, culture supernatant and process intermediate each interfere differently, and an assay developed in buffer will meet none of those challenges. Supply real samples, including the difficult ones, at the start of development rather than at the end.
3. **Screen reagent pairs rather than choosing one**: Capture and detection reagents that each work alone frequently fail together because they compete for the same epitope. Pair screening is a real experiment with a real cost, and a development plan without it is gambling on a first choice.
4. **Fix the standard and its traceability early**: The number the assay produces is relative to its standard. Recombinant material, a pooled natural sample and an international reference each give different absolute values. Choose one, document its source and lot, and plan how it will be replaced when it runs out.
5. **Establish the validation package the use requires**: Accuracy, precision within and between runs, selectivity, dilutional linearity, stability under the intended storage and handling, and a stated range. Regulated uses add more. Agreeing the package before development starts prevents an assay that works but cannot be signed off.
6. **Plan the transfer while the developer is still available**: A written method, a reagent list with lots, the standard curve model and acceptance criteria, plus a demonstration in the receiving laboratory's hands. Transfers arranged after the project closes tend not to happen.

**Why does an assay that works in buffer fail in serum?**

Matrix components bind reagents, alter the background and shift the curve. It is the single most common failure mode, and it is why real samples belong in development from the first screening rather than in a validation at the end.

**What should a development package contain at handover?**

The written method, the reagent identities and lots, the standard and its provenance, the curve fitting model, the acceptance criteria, and the raw data behind every validation claim. A report summarising results without the underlying data is not transferable.

**Is a multiplex format worth the added complexity?**

When several analytes genuinely inform one decision and sample volume is limited, yes. When one analyte matters, multiplexing adds cross reactivity and calibration problems for no benefit and makes troubleshooting considerably harder.

**How long does a custom development take?**

Longer than most plans, because reagent pair screening and matrix work are iterative. Compressed schedules usually compress validation, which is precisely the part that determines whether the assay can be used for anything important.

## Immune cell markers

Marker lists are easy to find and easy to misuse. A marker identifies a cell only within a gating context, and several of the most commonly used ones change with activation, with enzymatic dissociation or with an hour on the bench. Panels built from a list rather than from a gating strategy produce populations that are real on the plot and absent in the tissue. This page is about building the strategy first.

1. **Write the gating strategy before choosing fluorophores**: Decide the order of exclusions and inclusions on paper: debris, doublets, dead cells, lineage exclusions, then the positive definitions. The strategy determines which markers must be bright and which can tolerate a dim channel, and that is what should drive fluorophore assignment.
2. **Put the dim and the rare on the bright detectors**: Antigens expressed at low density and populations present at low frequency need the brightest available fluorophores on the cleanest detectors. Highly expressed lineage markers can sit anywhere. Getting this backwards is the most common reason a panel underperforms its design.
3. **Control for what handling does to the marker**: Several activation markers appear within minutes of manipulation and some surface antigens are cleaved by the enzymes used to dissociate tissue. Fix the handling protocol, keep it identical across groups, and where possible confirm with an antibody clone whose epitope survives the treatment.
4. **Exclude before you include**: Populations defined by the absence of other lineages need those exclusion channels to be reliable, because a leaky dump channel contaminates the population of interest with whatever it failed to remove. Spend detector quality on exclusions, not only on the markers of interest.
5. **Use full minus one controls, not isotypes**: Where a boundary must be drawn, a control lacking one color from the full panel shows where spread from every other channel puts the edge. Isotype controls do not address spillover spread and routinely place gates in the wrong position.
6. **Record clone, lot and titration with the protocol**: Two antibodies against the same target with different clones are different reagents. Titrate each on your own cells, record the titre, and treat a clone change as a protocol change requiring re titration.

**Why do monocyte markers define different cells in different papers?**

Because a marker is only interpretable inside a gating strategy. The same antigen appears on several lineages, and what identifies a population is the combination and the exclusions, not any single channel.

**Can a published panel be used as is?**

As a starting point only. Instrument configuration, fluorophore availability, sample type and handling all differ, and every antibody still needs titrating on your own cells. Published panels transfer the logic, not the settings.

**What about markers that dissociation destroys?**

Check whether the target is sensitive to the enzyme in use, and if it is, either change the dissociation method, choose a clone against a resistant epitope, or stop treating that marker as quantitative in dissociated tissue.

**How many colours does a panel really need?**

As many as the exclusions and subsets require and no more. Every added color adds spread to the others, so a well designed panel with fewer parameters frequently resolves rare populations better than a larger one designed by accumulation.

## Recombinant proteins

Expressing a protein is rarely the hard part. Getting it folded, getting it pure enough for the assay that will use it, and being able to say what you have are the parts that consume the time. The decisions that determine all three are made early, when the host and the construct are chosen, and they are expensive to revisit. This page walks them in that order.

1. **Start from what the protein has to do**: A binding reagent, a structural target, an enzyme assay substrate and an immunogen impose completely different requirements on folding, modification, purity and quantity. Write those requirements down first, because they select the host rather than the other way round.
2. **Choose the host from the modifications required**: Bacterial expression is fastest and cheapest and gives no glycosylation and limited disulphide formation in the cytoplasm. Insect systems handle larger, more complex proteins with some modification. Mammalian expression gives human like processing and is the default for secreted and glycosylated products. Cell free systems suit toxic proteins and very fast turnarounds at small scale.
3. **Design the tag and its removal together**: An affinity tag makes purification simple and can change solubility, activity and crystallisation. Decide at design time whether it stays or goes, and if it goes, include the protease site and plan the second chromatography step that removes the protease and the cleaved tag.
4. **Screen constructs and conditions in parallel**: Boundaries, tag position, fusion partners, temperature and induction strength all change yield and solubility, and small parallel screens answer in a week what serial optimisation answers in a quarter. This is the single highest return activity in the whole workflow.
5. **Build the purification around the contaminant that matters**: Affinity capture is a start, not a purification. Add polishing steps chosen against what the assay cannot tolerate: aggregates, endotoxin, host proteins, nucleic acid or the tag itself. A single step preparation is fine for a pull down and inadequate for a potency assay.
6. **Characterise before anyone uses the material**: Identity, concentration by a method appropriate to the protein, purity, aggregation state and activity. Report all five with the lot. A protein delivered with a concentration and nothing else will be argued about for months.

**Why is my protein in inclusion bodies?**

Usually expression faster than folding. Lower the temperature, reduce induction strength, try a solubility partner or move to a periplasmic or secreted route. Refolding is possible but adds a development project and a yield loss, so change the expression first.

**Should the affinity tag be removed?**

Remove it when the protein is used for structure, for activity measurements where the tag could interfere, or as an immunogen where the tag would dominate the response. Keep it for routine pull downs and immobilisation, where it is doing useful work.

**How should concentration be measured?**

By a method suited to the protein and stated with the result. Absorbance using a calculated extinction coefficient is accurate for proteins with aromatic residues and unreliable for those without; colorimetric assays are matrix sensitive and standard dependent. Say which was used.

**When is buying the protein the better answer?**

Whenever a catalog product exists at the required purity and activity, and the quantity is modest. Expression is worth doing when the variant is unusual, the quantity is large, the material must be owned, or no supplier offers the form you need.

## qPCR instruments and reagents

A quantitative amplification result is a number produced by an instrument, a chemistry, a primer design and a set of controls acting together, and weakness in any one of them is invisible in the output. Instruments are usually chosen on throughput and speed, which are the least important properties. This page sets out what actually determines whether the numbers are trustworthy.

1. **Check block uniformity before anything else**: If wells at the edge of the block run at a different temperature from the center, every plate carries a position effect that looks like biology. Ask for uniformity data and, better, run a uniform plate of the same sample across all wells during evaluation.
2. **Match the optics to the dyes you will actually use**: Excitation and emission channels determine which reporter and quencher combinations work and how many targets can be multiplexed. A platform with an attractive channel count that omits the dye your assays are built on is an expensive migration.
3. **Choose intercalating dye or probe chemistry deliberately**: A dye based mix is cheap and flexible and reports any double stranded product including primer artefacts, which is why a melt analysis is mandatory with it. Probe chemistry adds specificity and multiplexing at a higher per reaction cost. Decide per assay rather than per laboratory.
4. **Design primers against the target's structure**: Span an intron where genomic contamination is possible, check for secondary structure and for annealing to related sequences, and validate with a dilution series that gives an efficiency inside an acceptable band. Primers that were never validated are the commonest cause of an unreproducible result.
5. **Decide one step against two step for RNA targets**: One step reduces handling and contamination risk and suits routine panels. Two step lets one reverse transcription serve many targets and gives a stable complementary DNA stock, which suits exploratory work and precious samples.
6. **Write the control set into the protocol**: No template controls on every plate, no reverse transcription controls where RNA is the input, an inter plate calibrator where results are compared across runs, and reference targets chosen by stability in your own samples rather than by tradition.

**For a qPCR machine, does the instrument or the chemistry matter more?**

The chemistry and the primer design dominate result quality; the instrument mostly contributes block uniformity and optical sensitivity. A well designed assay runs acceptably on most modern instruments, and a poorly designed one fails on all of them.

**How should reference targets be chosen?**

Empirically, by measuring candidate references across your own experimental conditions and selecting those that do not change. Using a conventional housekeeping target without checking it is a frequent and invisible source of error.

**Is a melt curve always necessary?**

With intercalating dye chemistry, yes, because the dye reports any double stranded product including primer dimers. With probe chemistry it is unnecessary for specificity but remains a useful diagnostic during assay development.

**When is a gradient pcr machine still useful?**

During assay development, where a temperature gradient across the block finds the annealing optimum in one run. For routine running of validated assays it adds nothing, which is why many laboratories keep one older gradient instrument alongside their production platform.

## ELN and LIMS software

The notebook and the management system solve different problems, and most failed implementations start with a requirement that mixes them. A notebook records what a scientist did and thought. A management system moves samples through defined workflows and produces results against specifications. Buying one to do the other's job is the most expensive mistake in laboratory software, and it is entirely avoidable.

1. **Classify your work as exploratory or repetitive**: Work that differs every time, where the record is a narrative and an attachment, belongs in a notebook. Work that repeats against a defined procedure with a specification and a result belongs in a management system. Most organizations have both and should expect to buy both, at different times.
2. **Map the sample journey and the decision points**: Where does a sample enter, what states does it pass through, who approves each transition, and what record is made at each one? That map is the configuration requirement. Buying before drawing it is how implementations become eighteen month projects.
3. **Decide the deployment model on validation, not on price**: Hosted delivery removes infrastructure work and moves qualification responsibilities into supplier assessment and a defined update process. That is manageable, but it must be planned, because a vendor update arriving unannounced into a validated environment is a finding waiting to happen.
4. **Test configurability with a real workflow**: Ask the vendor to configure one of your actual workflows during evaluation, and note whether it was done by a consultant or by an administrator. Systems that only the supplier can change become frozen, and every subsequent requirement becomes a change order.
5. **Settle instrument integration early**: List the instruments whose data must arrive automatically, and ask what each connection costs and who maintains it when the instrument software updates. Integration is usually the largest and least visible line in the project.
6. **Plan the implementation as a project with an owner**: Data migration, master data cleanup, user training, parallel running and a cutover date. Name someone accountable and give them the time. Implementations in this category fail on ownership far more often than on software.

**Can one system be both notebook and management system?**

Several products offer both, and the combination works when one side is clearly dominant and the other is light. When both needs are substantial, a combined product usually does one of them poorly, and the interface between two good systems is easier to live with.

**Is hosted delivery acceptable in a regulated laboratory?**

Yes, routinely, provided the supplier is assessed, the responsibilities are documented in an agreement, and the update process is controlled so validated configurations are not changed without notice and reassessment.

**What does a free electronic lab notebook actually cost?**

Implementation, configuration, hosting, backup, support and the audit trail features that are usually reserved for paid tiers. For a small group tracking simple work it can be genuinely economical; for anything regulated it rarely is.

**How long should an implementation take?**

Longer than the sales cycle suggests and shorter than a failing project runs. The determining factors are how clean the existing data is, how many instrument integrations are in scope, and whether a named person owns the rollout.

**Is an electronic lab notebook free tier good enough?**

Often, for an individual or a small group, and a product people actually use beats a better one they do not. Check the export first: entries and attachments have to come out in a readable form if the tier changes.

## Plate washers and readers

Washing is where most immunoassay variability is created and where almost no purchasing attention is spent. A washer that leaves different residual volumes across the plate produces an edge effect that will be blamed on the reagents, the operator or the plate for months. This page is about the properties that actually decide whether a washer improves an assay or quietly degrades it.

1. **Ask for residual volume, measured**: The volume left in a well after aspiration is the property that carries reagent between steps. Ask for the figure, ask how it was measured, and ask for the variation across the plate rather than the best well.
2. **Check manifold design and blockage handling**: Narrow aspiration pins block, and a blocked pin produces one column of anomalous results that looks like a pipetting error. Look for a manifold that can be cleaned without tools and software that detects a failed aspiration rather than continuing silently.
3. **Confirm plate format and adherent cell compatibility**: Strip plates, deep well plates and cell culture plates each need different head geometry and dispense force. If cell based assays are in scope, a gentle dispense mode and adjustable pin height are not optional features.
4. **Plan the maintenance regime before purchase**: Daily rinse, weekly decontamination and a defined unblocking procedure keep a washer working. A washer without an owner becomes the source of the drift it was bought to remove.

**Does a washer actually reduce variability?**

A well specified and maintained one does, because it removes operator to operator differences in soak time and aspiration. A poorly maintained one increases variability, because blocked pins create systematic position effects that manual washing would not produce.

**How many wash cycles are needed?**

Enough to reduce background to a stable level, determined experimentally for the assay rather than taken from a kit insert. Additional cycles beyond that point cost time and, in cell based assays, cost cells.

**Can one ELISA plate washer handle cell based and plate based assays?**

Only if it offers a gentle dispense mode and adjustable pin height. A washer set up for robust immunoassay plates will lift adherent cells, and the resulting data looks like a biological effect.

## Manual pipettes and tips

A pipette is a calibrated instrument and a tip is part of the measurement, not a consumable detail. Changing tip brand without re verifying volume is a silent method change, and it is the most common unrecorded source of drift in a laboratory. This page covers what to specify on the pipette, what to specify on the tip, and how to keep the pair delivering what the protocol says.

1. **Choose the volume range tightly**: Accuracy is worst at the bottom of a pipette's range, so a volume routinely dispensed near the minimum belongs on a smaller instrument. Covering the laboratory's real volumes with two or three well chosen ranges beats one general purpose instrument used everywhere.
2. **Treat the tip as part of the instrument**: Fit, sealing force and the geometry of the tip cone all affect the delivered volume. If the laboratory changes tip supplier, verify the volume with the new tip before the change is adopted, and record the verification.
3. **Decide where filters are genuinely required**: Filter tips protect the pipette and the sample from aerosol carryover, which matters for amplification work, for cell culture and for anything infectious. Elsewhere they add cost and slightly reduce the usable volume.
4. **Match the orifice to the liquid**: Wide bore tips protect fragile cells and handle viscous liquids without shear; positive displacement tips handle volatile and dense liquids that defeat an air cushion. Using a standard tip on either produces a systematic volume error nobody sees.
5. **Set a calibration interval and keep the certificates**: Interval by use rather than by calendar alone, with certificates stating the tolerances tested, the tips used and the environmental conditions. A certificate that does not name the tip used is describing a different instrument from the one on your bench.

**Do third party tips really change the result?**

Yes, measurably. Fit and sealing force vary between manufacturers, and the effect is largest at low volumes where tolerances are tightest. Universal fit is a claim about geometry, not about equivalence of delivered volume.

**How often should a pipette be calibrated?**

Set the interval from usage and consequence: heavily used instruments feeding critical results need more frequent checks than occasional ones. Add a simple in house gravimetric check between formal calibrations to catch drift early.

**Is an electronic pipette worth it?**

For repetitive dispensing, long series and users with repetitive strain concerns, yes, because it removes plunger force variation between operators. For occasional general use it adds cost, charging and another item to maintain.

**Does a p200 from one maker match a p200 from another?**

Not necessarily. The naming is a convention and the ranges overlap differently between makers, particularly at the lower limit. Read the stated range and match it to the volumes your protocols actually use.

## Process analysers and sensors

An inline analyser is bought as hardware and lives as a model. The spectrometer is the easy part; the calibration that turns a spectrum into a concentration is built from reference samples covering the whole operating range, and it has to be maintained as the process changes. Programs that budget for the instrument and not for the model are the ones where the probe ends up unused.

1. **Define the measurement and its required accuracy**: What decision does the inline number support, and how accurate must it be for that decision? An analyser accurate enough to see a trend is far cheaper than one accurate enough to release material, and the two are frequently confused at specification.
2. **Choose the probe position before the instrument**: Path length, window fouling, bubbles and temperature at the measurement point determine whether any spectrum is usable. A probe in a poorly chosen position produces uninterpretable data regardless of the instrument behind it.
3. **Plan the reference data set honestly**: A calibration model needs samples spanning the full range of the variable, including conditions the process rarely reaches. Collecting them may require deliberately running the process off target, and that has to be planned and approved.
4. **Decide how the model will be maintained**: Raw materials, process changes and probe replacement all shift the spectral response. Agree who owns the model, how drift is detected and what triggers recalibration, before the system is handed over.
5. **Qualify the sensor against a reference method**: Every inline sensor needs an offline reference to be judged against, at installation and periodically afterwards. A sensor that has never been compared to a reference method is a trend line, not a measurement.

**Why do inline analysers end up unused?**

Almost always because the calibration model was never completed or maintained. The hardware works; the model that turns its output into a usable number requires reference data and ownership that the project did not budget.

**Is Raman or near infrared the right choice?**

Raman handles aqueous systems well and gives sharper molecular specificity, which suits identification and some concentration work. Near infrared is faster, cheaper and well suited to bulk composition and moisture. The decision follows the analyte and the matrix, not the technology.

**Are single use sensors, such as an optical dissolved oxygen sensor, as good as traditional probes?**

For dissolved oxygen and pressure they are now routinely used in disposable systems, with pre calibration replacing in place calibration. The trade is convenience and sterility against the inability to recalibrate in place, and the supply chain becomes part of the measurement.

**What does a handheld instrument actually deliver?**

Identification and verification of a material against a library, quickly and at the point of receipt. Quantitative work in a process stream is a different specification and usually a different instrument.

**Does an optical composition instrument measure protein directly?**

No. It predicts the value from a calibration model built against a reference method. Ask which reference method, how many samples, whether your material is inside the model's range and how the model is maintained.

**Why is a single oil analysis result of limited use?**

Because condition monitoring is about trend: wear metals rising, water ingress, viscosity changing. Consistent sampling points and a history matter more than instrument precision.

**Are catalogue food analysers all the same technology?**

No. The heading mixes optical composition instruments, classical nitrogen and fat methods, moisture balances and color meters. The specification your product is sold against names the reference method the result has to match.

## Bioinformatics and library prep

The sequencing is rarely the limiting step. What the library preparation captured, how samples were batched, and whether the analysis can be rerun by someone else determine whether the data answers the question. Buying analysis separately from preparation, and specifying neither in detail, is how projects arrive at a folder of figures nobody can reproduce.

1. **Start from the question and the effect size**: Detection of a rare variant, quantification of a modest expression change and discovery of a novel transcript need different coverage, different replication and sometimes different chemistry. The statistical requirement sets the design, and it should be agreed before samples are prepared.
2. **Choose the library chemistry against the molecule**: Short fragmented inputs, degraded material, low input samples and small RNA each need their own preparation. A kit chosen for convenience rather than for the input type is the most common reason a run produces libraries that do not represent the sample.
3. **Randomise across batches deliberately**: Preparation batch, index set and flow cell lane all leave signatures. Distribute experimental groups across batches rather than processing one group at a time, and record the batch structure so it can be modelled later.
4. **Agree the deliverables in writing**: Raw reads, alignment files, the processing scripts with versions and parameters, the reference used, quality metrics and the intermediate outputs. A report with figures and no underlying data is not a deliverable, it is a summary.
5. **Confirm data retention and transfer before starting**: Where the data lives, for how long, who pays for storage after the project and how it is delivered. Large data sets arriving without a transfer plan sit on a provider's server until the invoice for storage arrives.

**What should a bioinformatics deliverable include?**

Raw and aligned data, the exact pipeline with software versions and parameters, the reference genome and annotation used, quality control metrics for every sample, and the code that produced each figure. Anything less cannot be reproduced or extended.

**Is a targeted panel better than whole exome or genome?**

Where the regions of interest are known, a panel gives deeper coverage for the same cost and a smaller analysis burden. Where they are not, broader sequencing avoids the ascertainment bias of having chosen the targets in advance.

**Does NGS library preparation need automation?**

Above a few dozen samples, automation reduces the batch to batch variation that manual preparation introduces and makes the batch structure predictable. Below that, careful manual work with deliberate randomisation is usually sufficient.

**Who should own the analysis in the long run?**

Whoever will be asked to defend it. Outsourced analysis is efficient for well defined pipelines; exploratory work that will be iterated benefits from in house capability, or at least from a provider who hands over runnable code.

## PCR and gel reagents

Ladders, polymerases and master mixes are ordered by habit and treated as interchangeable, and they are not. Fidelity differs by orders of magnitude between enzymes, master mixes differ in what they already contain, and a ladder chosen for the wrong size range makes a gel harder to read rather than easier. This page is about choosing them on purpose.

1. **Match the ladder to the range you are resolving**: A ladder is a ruler, and a ruler with marks far from your fragment sizes is a poor one. Pick the ladder whose bands bracket the expected product closely, and prefer one with a reference band of distinct intensity so orientation is obvious.
2. **Choose the polymerase by what happens to the product**: A product that will be sequenced, cloned and expressed needs a proofreading enzyme, because errors propagate into everything built from it. A product that will be looked at on a gel and discarded does not. The cost difference is trivial next to the cost of an error in a construct.
3. **Use a hot start enzyme for anything multiplexed or sensitive**: Activity at room temperature during setup generates primer dimers and non specific products before cycling begins. A hot start format removes that window and is close to free at the scale most laboratories work.
4. **Read what the master mix already contains**: Mixes differ in magnesium concentration, in whether a passive reference dye is included and in additives that help difficult templates. Adding what is already there, or assuming a dye is present when it is not, produces failures that look like template problems.
5. **Control the plate medium if counts are compared**: Batch, pour depth, drying and age all change colony size and recovery. Where counts are compared across time, standardize preparation or buy pre poured plates from one supplier with a stated lot.

**Does Taq DNA polymerase fidelity matter for routine PCR?**

It matters whenever the product is kept. For screening and genotyping, where the product is read and discarded, a standard enzyme is fine. For cloning, sequencing and library preparation, a proofreading enzyme avoids errors that become permanent.

**Why does my 1 kb DNA ladder look uneven?**

Usually loading or running conditions rather than the ladder itself: too much material, a gel percentage unsuited to the range, or a run that went too far. Ladders are supplied at concentrations assuming a defined gel volume, and overloading distorts the smaller bands.

**Are master mixes worth the premium?**

Almost always, because they remove pipetting steps and the variability that comes with them, and because lot to lot consistency is the manufacturer's problem rather than yours. Mixing from components makes sense only at high volume or where a component must be varied.

**How often should mycoplasma testing be run?**

On arrival for every new line, after any quarantine period, and on a routine schedule for lines in continuous culture. Testing only when something looks wrong means the contaminated cultures have already been used.

## Benchtop lab equipment

Benchtop equipment is bought on price because every model appears to do the same thing. What separates them is temperature uniformity over a full load, how quickly a unit recovers after the lid opens, and how easily it can be cleaned when something grows in it. Those properties are invisible on a specification sheet and obvious six months in.

1. **Specify uniformity under load, not empty**: A bath or incubator holding its setpoint empty says little. Ask for uniformity and recovery figures with a representative load, because that is the condition every real experiment runs in.
2. **Plan the cleaning regime before purchase**: Water baths grow biofilm, incubators grow fungus, and both contaminate cultures quietly. Choose units that can be fully drained and wiped, prefer smooth interiors without inaccessible corners, and write the cleaning schedule into the purchase decision.
3. **Separate refrigerated from ambient centrifugation**: Temperature sensitive preparations need a refrigerated rotor and the cooling time that goes with it; routine spins do not. Buying one refrigerated unit for a group and a simple high speed unit per bench is usually better value than a compromise on both benches.
4. **Use a dedicated hood for amplification setup**: An enclosure with ultraviolet decontamination and clean airflow for reaction setup protects against amplified product carryover. It is not a biosafety cabinet and does not protect the operator, and using one for the other's job is a real hazard.
5. **Standardize consumables that touch every protocol**: Tube sizes, rack formats and rotor adaptors propagate through every method in the laboratory. Settle them once so that a protocol written at one bench runs at another without substitution.

**Is a water bath still necessary?**

For gentle, uniform heating of liquids, yes, and it remains the most reliable way to bring a large volume to temperature. Dry block heaters are better for small tubes and far cleaner, which is why most laboratories end up with both.

**What causes contamination in a cell culture incubator?**

Usually the water tray and the interior surfaces rather than the air. Regular emptying and cleaning of the tray, a suitable additive where the manufacturer permits one, and a defined decontamination schedule prevent nearly all of it.

**Does a PCR hood replace a biosafety cabinet?**

No. A setup enclosure protects the reaction from contamination; a biosafety cabinet protects the operator and the environment from the material. They serve different purposes and containment requirements are decided by the material, not by convenience.

**How much does shaker capacity really matter?**

More than expected, because capacity constrains experiment size and because a shaker running at its limit wears quickly. Size for the largest routine experiment rather than the average one, and check the platform accepts the flask sizes in use.

**When is a shaking water bath the better instrument?**

Whenever the contents need mixing as well as warming: a hybridisation, an enzymatic digest of a viscous sample, or a small culture where an air shaker would dry the vessel. The trade is that the mechanism sits over water, so it corrodes, and the bath has to be drained and dried rather than topped up. For simple warming a still bath lasts longer and costs less.

## Live cell imaging

A time lapse experiment fails for boring reasons: the cells drifted out of focus overnight, the stage moved, the environment was not stable, or the illumination that produced beautiful images also killed the cells slowly enough that the death looked like biology. Optical performance is rarely the binding constraint. This page is about the constraints that actually bind.

1. **Budget light as if it were a reagent**: Every exposure costs the cells something. Decide the minimum signal that answers the question and work down to it with sensitive detection, longer intervals and lower intensity rather than up from a comfortable image. A visibly healthy culture at the end of a run is part of the result.
2. **Specify the environment as a system, not an accessory**: Temperature, carbon dioxide and humidity must be stable at the sample over the full experiment, not at a controller somewhere nearby. Stage top chambers and full enclosures behave differently, and condensation on the optical path ends runs that were otherwise fine.
3. **Test autofocus over the real duration**: Focus drift over hours is the commonest cause of an unusable time lapse. Hardware based focus holding is far more reliable than software refocusing, and the only meaningful test is an overnight run on the candidate system with your own vessel.
4. **Match the format to the question**: A few dishes followed closely is a microscope problem. Many wells sampled repeatedly is a screening problem, and the instruments, the analysis and the cost structure differ completely. Choosing the wrong one means either wasted capability or an experiment that cannot be run.
5. **Plan storage and analysis before the first run**: Multi position, multi channel time lapse generates data faster than most laboratories expect, and an analysis pipeline arranged after acquisition usually means reacquiring. Decide the storage location, the naming convention and the analysis route in advance.

**How do I tell phototoxicity from biology?**

Run an identical experiment at lower illumination and longer intervals, and compare. If the effect weakens as light dose falls, the light was causing it. Including an unimaged control well from the same plate is the simplest version of this check.

**Is an incubator mounted imager enough?**

For growth, confluence, migration and simple fluorescent readouts across many wells, frequently yes, and the environmental stability is excellent because the instrument sits inside the incubator. For high resolution subcellular work it is not a substitute for a microscope.

**What makes high content screening instruments expensive?**

The analysis and the data management rather than the instrument. Image analysis development, storage and the people who maintain both usually exceed the hardware cost over the instrument's life, and they are routinely omitted from the business case.

**Which vessel should be used?**

Whichever has an optical bottom matched to the objective's correction, held flat and stable on the stage. Vessel bottom thickness variation is a real and underappreciated source of focus and aberration problems in long runs.

## Gene synthesis

Synthesis has become cheap enough that building a construct by hand is often the more expensive option, but the price per base hides the part that matters: whether your particular sequence is easy or hard to make. Repeats, extreme composition and structure turn a routine order into a quotation with a long lead time or a refusal. This page is about anticipating that before the order.

1. **Screen the sequence for difficulty before ordering**: Long repeats, homopolymer runs, extreme composition and strong secondary structure are what make synthesis slow or impossible. Suppliers screen automatically and will tell you, but finding out during quotation rather than after a delay is worth doing deliberately.
2. **Decide fragment or finished construct**: A linear fragment is cheaper and leaves you the assembly. A sequence verified plasmid arrives ready to use and includes the cloning and the verification. Price the bench time before choosing the fragment, because the assembly is rarely free.
3. **Choose codon adaptation with a reason**: Adapting codons to the expression host can raise yield substantially, and it can also remove regulatory elements, change folding kinetics or break an intended sequence feature. Adapt when expression is the goal, keep the native sequence when the sequence itself is under study.
4. **Specify the delivery format and the ends**: Vector, insertion site, tags, overhangs and the exact junction sequences belong in the order, not in a follow up email. Most rework in this category comes from an assumed backbone or an assumed reading frame.
5. **Keep the verification evidence with the construct**: Ask for the full sequence as manufactured and the underlying reads. File them against the construct identifier so the next person does not resequence a construct that was already verified.

**Why was my sequence refused or delayed?**

Almost always structural difficulty: long repeats, extreme composition, homopolymer runs or strong secondary structure. Splitting the sequence, silently varying codons in the difficult region or accepting a longer lead time are the usual remedies.

**Should I always codon optimise?**

No. Optimize when the aim is protein yield in a known host. Keep the native sequence when regulatory elements, splicing, translational pausing or the sequence itself is part of what you are studying, because adaptation quietly removes those features.

**Is a fragment cheaper than a clone?**

Per base, yes. In total, frequently not, once assembly, transformation, screening and sequence verification are costed as labor. Fragments make sense when you are assembling several of them into something a supplier could not deliver anyway.

**How does synthesis compare with amplifying from cDNA?**

Amplification gives you the natural sequence including any variant present in your source, which is sometimes exactly what you want and sometimes a surprise. Synthesis gives you the sequence you specified, which is easier to defend and to reproduce.

**Is artificial gene synthesis cheaper than amplifying and cloning?**

For anything you do not already hold, usually yes once a person's time is counted: a fragment arrives sequence perfect and assembles in an afternoon, where a clone from template needs primers, a reaction that may not work, a gel and a sequencing round. Where the template is already in the freezer and the construct is simple, amplification is still the cheaper route.

## Particle and thermal analysis

A particle size result is the product of a sample preparation, a dispersion, an optical model and a choice of how to report the distribution. Change any of those and the number changes, which is why two laboratories measuring the same powder routinely disagree. Buying the instrument is the easy half; specifying the method is the half that determines whether the data means anything.

1. **Decide wet or dry dispersion from the material**: Dry dispersion suits free flowing powders and avoids solvent compatibility questions, but the air pressure that separates agglomerates can also break fragile particles. Wet dispersion is gentler and needs a liquid the material neither dissolves in nor swells in. Many materials need both, measured and compared.
2. **Establish the optical parameters honestly**: The model used to convert scattering into a size distribution needs the refractive index of the material and the medium. Guessing these produces a plausible distribution that is systematically wrong, particularly at the fine end.
3. **Fix how the distribution is reported**: Volume, surface and number based distributions of the same sample look completely different, and percentile descriptors depend on which basis is used. Agree the basis and the descriptors before any specification is written against them.
4. **Demonstrate dispersion adequacy, not just repeatability**: A result that repeats precisely may be repeatably measuring agglomerates. Vary the dispersion energy and look for a plateau where the result stops changing. That plateau, not the repeatability, shows the particles are dispersed.
5. **Qualify with a reference material**: A certified size standard run periodically shows the instrument and the method are still behaving. It is the only external check available, and it belongs in the routine rather than in the installation only.

**Why do two laboratories get different sizes for the same powder?**

Almost always different dispersion or different optical parameters rather than different instruments. Harmonising the method, including the dispersion energy and the refractive index values, resolves most disagreements of this kind.

**Which distribution basis should be quoted?**

Whichever the specification was written against, stated explicitly. Volume based descriptors are conventional for many materials, but quoting a percentile without saying which basis produced it is meaningless.

**Is diffraction the right technique for fine particles?**

It covers a wide range but loses discrimination at the fine end where other techniques are stronger. If the material of interest sits mostly below the comfortable range, a complementary technique is needed rather than a more expensive diffraction instrument.

**Where does thermal analysis fit?**

It answers a different question: transitions, decomposition and stability rather than size. Laboratories characterising formulated materials usually need both, and they are separate purchases with separate method development.

## Aseptic filling and monitoring

Filling equipment is quoted on units per hour and judged on something else entirely: how many human interventions the process requires, whether the monitoring can show the environment held, and whether the records produced would survive scrutiny. A fast machine that needs frequent operator access into the critical zone is a slower machine once the rejects are counted.

1. **Count the interventions, not the units per hour**: Every entry into the critical zone is a contamination risk and has to be justified, simulated and recorded. Machines differ enormously in how many routine and corrective interventions they demand, and that number predicts batch outcomes better than nominal throughput.
2. **Decide the barrier approach early**: An open line in a classified room, a restricted access barrier and an isolator impose very different facility, gowning and monitoring requirements. This choice shapes the building, so it belongs at the start rather than after equipment selection.
3. **Design monitoring to answer a question**: Sample locations should be chosen by risk, and the reason for each location recorded. Viable and non viable sampling, personnel monitoring and surface sampling each answer something different, and a plan assembled from convenience rather than from risk produces data nobody can interpret.
4. **Prove the process with a media fill**: A process simulation running the full process with growth medium, including the worst case interventions and the longest intended duration, is what demonstrates the line as operated. A simulation that omits the difficult interventions demonstrates a process you do not run.
5. **Assess data integrity across the whole record chain**: Instrument outputs, batch records, environmental monitoring results and the systems holding them each need attributable, legible, contemporaneous, original and accurate records. Assess the chain end to end, because the weakest link is usually a standalone instrument nobody listed.
6. **Qualify format changes as changes**: A new container, closure or fill volume is a change to a validated process, with its own qualification and often its own simulation. Lines bought for flexibility still require that work for each format.

**Does an isolator remove the need for environmental monitoring?**

No. It changes what is monitored and how often, and it usually reduces personnel related risk substantially, but the requirement to demonstrate that the environment held during the process remains.

**What makes a media fill meaningful?**

That it reflects the process as actually operated, including the interventions, the duration, the shift pattern and the worst case conditions. A simulation designed to pass rather than to challenge the process is evidence of nothing.

**Where do data integrity findings usually come from?**

Standalone instruments with local storage, shared logins, records that can be overwritten, and results that exist only as printouts. The laboratory rather than the production line is where these are most often found.

**Can a laboratory scale filler be scaled up later?**

Sometimes mechanically, rarely without requalification. Treat a small scale filler as suitable for its own scale, and plan the larger line as a separate project with its own qualification rather than as an upgrade.

## Primary human cells

Primary cells are the reason an experiment resembles the biology and the reason it is hard to repeat. Every donor is a different lot, every isolation method leaves a different population, and the material arrives with obligations about consent and handling that a cell line never carried. This page is about designing around all three rather than discovering them mid study.

1. **Settle provenance and consent before sourcing**: Material from people comes with a consent scope, an ethics approval and a chain of custody. Establish what the consent permits, who holds the approval covering your use, and what the supplier can document, because a study built on material whose consent does not cover the work cannot be published or used.
2. **Choose fresh or cryopreserved deliberately**: Fresh material gives the closest thing to the in vivo state and forces the experiment to run on the supplier's schedule. Cryopreserved vials give control of timing and lose a fraction of viability and some functional responses. Neither is better; the choice should follow whether the readout is sensitive to what freezing changes.
3. **Fix the isolation method and keep it fixed**: Density separation, negative selection and positive selection each leave a different population, and positive selection leaves an antibody bound to the marker used. Comparing across methods compares methods. Choose one, document it, and note it in every result.
4. **Plan for donor variation in the design**: Donor to donor differences frequently exceed the effect under study. Use several donors rather than several wells from one, treat donor as a blocking factor in the analysis, and record donor characteristics that could plausibly matter.
5. **Match medium, coating and passage limit to the cell**: Primary cells have finite lifespans and change with passage. Establish the passage limit at which the phenotype still holds, write it into the protocol, and match the coating and medium to the cell type rather than to whatever the laboratory has.
6. **Handle at the containment the material requires**: Human derived material is handled as potentially infectious, with the precautions that implies for the people doing it. Confirm the containment level, the waste route and the vaccination requirements before the first delivery.

**How many donors does an experiment need?**

Enough that a conclusion is not a property of one person. Three is a common minimum for exploratory work and frequently too few when variation is large; the honest answer comes from the variation observed in a pilot rather than from convention.

**Does positive selection change the cells?**

It can. An antibody bound to a surface marker may signal through it, and residual beads affect downstream steps. Negative selection avoids this at the cost of lower purity, which is the trade to make consciously.

**Why do results differ between vial lots?**

Because a lot is a donor. Differences in genotype, age, health, collection and processing all travel with the vial. Buying several vials from one donor lot lets a study run without confounding donor with experimental group.

**What documentation should arrive with the material?**

Donor identifier or code, consent scope, the collection and processing dates, the isolation method, viability, purity by a stated method, and any donor characteristics the supplier is permitted to share. Anything less makes the material hard to defend later.

**How long can I culture a primary cell?**

Less time than most protocols assume. Phenotype drifts with every division, so the useful window is early. Record donor, lot and passage, bank early, and use several donors where the conclusion is about people.

**Should I make my own stem-cell-derived model?**

Only if the genotype is the point and no characterised line exists. Reprogramming, characterisation, karyotype and differentiation development are a project; buying a characterised line or preparation is usually faster and cheaper.

## Cell analysers and sorters

Three instrument classes get discussed as one. A counter tells you how many cells and how many are alive. An analyser tells you what they express, one cell at a time. A sorter physically separates them, and in doing so becomes a containment problem. Buying up the chain because it seems safer is how laboratories acquire instruments nobody is trained to run.

1. **Decide whether cells must be recovered**: If the experiment ends with a measurement, an analyser suffices. If it continues with the cells, you need a sorter, and that changes the room, the training, the maintenance and the cost by a large factor. This is the first question and it settles most of the decision.
2. **Count the parameters the panel genuinely needs**: Detector count should follow from the panel design, including exclusions and viability, with a margin for the panel you will design next year. Buying maximum parameters for a four color laboratory wastes money and adds spread.
3. **Match lasers to the fluorophores, not the other way round**: The laser lines determine which dyes are excitable and how brightly. Check that the reagents your assays already depend on work well on the proposed configuration, because rebuilding validated panels around a new instrument is a hidden migration cost.
4. **Treat sorting as a biosafety project**: Sorters generate aerosols. Handling human or infectious material on one requires containment appropriate to the agent, verified airflow, and a documented failure procedure. Involve the biosafety officer at specification, not at installation.
5. **Plan quality control from day one**: Daily performance tracking with beads, recorded over time, is what tells you an instrument has drifted rather than the biology having changed. Instruments differ in how easy this is to run and to review, which matters more than a headline sensitivity figure.
6. **Budget the operator, not just the instrument**: Analysers can be shared and largely self service; sorters realistically need a trained operator. If no such person exists and none is being hired, a sorting service is cheaper and gives better data than an underused instrument.

**Can a cell analyzer be upgraded to a cell sorter?**

Generally no. Sorting requires fluidics, droplet formation and charging hardware that are designed in, and the containment requirements differ. Treat it as a different purchase rather than an option to add later.

**Is a benchtop flow cytometer enough?**

For most immunophenotyping with moderate parameter counts, yes, and the simplicity is a real advantage in a shared laboratory. The limits appear with very large panels, rare event detection and any requirement to recover cells.

**When is a simple counter the right purchase?**

When the question is concentration and viability for seeding, passaging or dosing. Using a full analyser for that ties up an expensive instrument and produces no additional information.

**Should sorting be outsourced?**

Frequently yes, particularly where sorting is occasional. A core facility or a service brings a trained operator, maintained instruments and the containment already in place, and the per sample cost is usually below the true cost of owning one.

**What does a cell separator do instead of a sorter?**

It enriches rather than selects: a magnetic or density based separator takes a bulk population out of a suspension by one property, in minutes and with no instrument time. It cannot select on several markers at once or recover a rare population purely, which is what a sorter is for. Most protocols use both, a separator to remove the bulk and a sorter to take what is left.

## Physical characterisation

Four small instruments answer four questions that people routinely conflate, and each reports a number whose meaning depends entirely on the method behind it. Two moisture analysers using different principles will disagree about the same sample, and both will be right about what they measured. This page separates the questions and says which instrument answers each.

1. **Decide what moisture means for your sample**: Loss on drying measures everything volatile at the chosen temperature; a specific chemical titration measures water alone. For a material that releases solvent or degrades on heating, the two differ substantially, and the specification has to name the method.
2. **Use osmolality where tonicity matters**: Osmolality is the property that decides whether a formulation is tolerated by cells or by a patient, and it is not calculable reliably from composition once excipients interact. Freezing point depression is the common method and it is fast, small volume and robust.
3. **Automate titration where the volume justifies it**: An automatic titrator removes endpoint judgement and produces a record. Below a few determinations a week the manual method is fine; above that the consistency and the audit trail usually pay for the instrument.
4. **Match the calorimetric technique to the question**: Thermal transitions, decomposition and interaction energetics are different measurements with different instruments. Buying a general purpose unit for a specific stability question frequently produces data that cannot answer it.
5. **Decide in house against a testing laboratory**: Occasional determinations, unusual methods and anything requiring accreditation are usually cheaper and more defensible bought in. Routine, high volume determinations belong in house where the method can be controlled and the turnaround is immediate.

**Why do two moisture results disagree?**

Because they measure different things. A drying method reports total volatile loss at a temperature; a chemical titration reports water specifically. On a sample containing other volatiles, the drying figure will be higher and neither is wrong.

**Can osmolality be calculated instead of measured?**

Approximately, and the approximation degrades as formulations get more complex. Where tonicity affects cells or patients, measurement is the defensible route and the instrument is inexpensive relative to the risk.

**What does an automatic titrator add?**

Consistent endpoint detection, unattended operation and a record of every determination. The chemistry is unchanged; what improves is reproducibility between operators and the ability to show how a number was obtained.

**Is calorimetry worth it for a small laboratory?**

Only with a recurring question that it answers. These instruments need method development and a trained user, and for occasional stability questions a contract laboratory delivers better data for less.

## Complement reagents

Complement is a cascade of proteolytic fragments, and almost every difficulty in measuring it comes from that fact. The protein you name is cleaved within minutes of a badly handled sample, the antibody you buy may recognize the parent, the fragment, the neoepitope or all three, and the result depends on which. This page is about choosing reagents that answer the intended question.

1. **Name the fragment, not the protein**: Detecting the intact component, a cleavage product or the neoepitope exposed by activation are three different measurements. Suppliers list them under the same protein name, so read the immunogen and the stated specificity rather than the product title.
2. **Fix sample collection before anything else**: Complement activates ex vivo. The anticoagulant, the time to processing and the temperature all change the result, and a sample handled casually reports activation that happened in the tube. Standardize collection, process quickly, and aliquot to avoid freeze thaw.
3. **Separate functional from quantitative questions**: A functional assay asks whether the pathway can proceed to completion; an immunoassay asks how much of a component is present. A deficiency shows in the first and may be invisible in the second, and the two are routinely confused when a result is interpreted.
4. **Choose the pathway you are interrogating**: The classical, lectin and alternative routes converge but are initiated differently, and an assay that isolates one requires conditions that suppress the others. Using a general activation readout and inferring the pathway is a common error.
5. **Include activation and inhibition controls**: A positive control that activates the cascade and a sample held under conditions that prevent activation bracket every run. Without them, a low result cannot be distinguished from a failed assay and a high one from a handling artefact.

**Why do complement results vary so much between laboratories?**

Sample handling, overwhelmingly. Time to processing, temperature and freeze thaw cycles change activation markers substantially, and these differences are rarely reported in enough detail to be reconciled.

**Does an antibody against a component detect its fragments?**

It depends entirely on the immunogen and the epitope. Some recognize the parent and every fragment containing that region; others are specific to a neoepitope that exists only after cleavage. The datasheet has to be read, not the product name.

**When is a functional assay necessary?**

Whenever the question is whether the pathway works, for example in suspected deficiency or when assessing an inhibitor. Component concentrations can look normal while function is absent, and only a functional readout distinguishes the two.

**How should serum samples be stored?**

Processed quickly, aliquoted in small volumes and held at low temperature, with the number of freeze thaw cycles recorded. Repeated thawing is the most common preventable cause of an uninterpretable complement result.

## Peptide synthesis

A peptide arrives as a white powder whose mass on the label is not the mass of peptide in the tube. Purity grade, counterion, water content and the analysis behind the certificate decide whether the material behaves as the sequence predicts, and they vary far more between suppliers than the price per residue does. This page is about specifying all of it.

1. **Set the purity grade against the use**: Crude material is adequate for screening a library of sequences. Purified material is required for quantitative assays, and high purity with full characterisation for anything going into animals. Ordering higher than needed wastes money; ordering lower produces results attributable to the impurities.
2. **Ask what the stated mass actually contains**: A quoted quantity is often gross weight including counterion and water, which can be a substantial fraction. Ask for net peptide content, and if the assay depends on concentration, ask for amino acid analysis rather than a nominal figure.
3. **Declare modifications and the counterion**: Terminal capping, labels, cyclisation, phosphorylation and unusual residues each change the synthesis and the price. The counterion matters too: a residue from purification can be biologically active in cell assays, so specify it where it could interfere.
4. **Flag difficult sequences early**: Long sequences, hydrophobic stretches, multiple cysteines and aggregation prone regions are what make a synthesis fail or a purification lose most of the material. Suppliers will tell you before quoting if asked, and a redesign at that stage is cheap.
5. **Specify the analytical package**: Mass confirmation and a purity chromatogram as a minimum, with amino acid analysis where quantity matters and a disulphide assignment where folding matters. The chromatogram should come with the material, not on request.
6. **Plan storage and handling on arrival**: Peptides are hygroscopic and many oxidise. Aliquot from a single dissolution rather than repeatedly opening a cold vial into a humid room, and record the solvent and concentration used.

**Why is my peptide less soluble than expected?**

Composition, usually. Hydrophobic and aggregation prone sequences need a solvent strategy rather than water, and the order in which solvents are added matters. Ask the supplier for a recommended dissolution before assuming the material is faulty.

**What does purity on a certificate mean?**

The area of the main peak in a stated chromatographic method, which says nothing about what the other peaks are or about water and counterion content. Two certificates quoting the same purity can describe materially different tubes.

**Is a peptide synthesis machine in house worth it?**

For groups making many sequences continuously, yes, because turnaround becomes hours rather than weeks. For occasional peptides it rarely is once instrument, reagents, purification and the trained person are counted.

**When is amino acid analysis needed?**

Whenever the result depends on knowing the concentration accurately, for example potency work or binding constants. Weighing a hygroscopic powder of unknown net content is not a quantitative operation.

**Why do two peptide synthesis price quotes differ so much?**

Length, purity and scale, plus any modification, cyclisation or label. Ask for the price against a stated purity and delivered quantity, and for what the certificate includes.

**Is purity the same as peptide content?**

No. Purity by chromatography is the fraction of peptide material that is your sequence; net peptide content is how much of the weighed powder is peptide rather than counter-ion and water. Dosing needs the second.

## Custom plasmid synthesis

Four adjacent products are ordered from the same suppliers and confused constantly: a linear fragment, a finished plasmid, a preparation service that grows and purifies a construct you already have, and a synthetic guide. Each solves a different part of the workflow, and ordering the wrong one wastes a week rather than money. This page separates them.

1. **Decide whether you need DNA or a vector**: A linear fragment is cheap and is an input to your own assembly. A finished plasmid is more expensive and arrives ready to transfect. If you have no assembly capacity or no time, the fragment is not the saving it appears to be.
2. **Specify the backbone and the junctions exactly**: Vector, insertion site, reading frame, tags and their positions, and the exact junction sequences. Most rework in this category is an assumed backbone or a frame shift introduced by an unstated junction.
3. **Send the sequence for screening before ordering**: Repeats, extreme composition and hairpins turn a routine synthesis into a long one or a refusal. Suppliers screen a draft for free, and a small redesign at that point avoids a delay measured in weeks.
4. **Use a preparation service where the bottleneck is quantity**: When the construct already exists and what you lack is milligrams at a stated purity, a preparation service is faster and cheaper than culturing it yourself, and it comes with a certificate you can file.
5. **Handle synthetic guides as their own class**: Chemically synthesised guides arrive with modifications that affect stability and immune sensing, and they are stored and resuspended differently from plasmid DNA. Order them with the modification pattern stated and keep them separate from the plasmid workflow.
6. **Settle ownership and sequence evidence in the order**: Ask for the full sequence as manufactured and the underlying reads, and confirm who owns the construct and any intermediates. Both are easy to agree before an order and awkward afterwards.

**Is a geneblock fragment enough, or do I need the finished plasmid?**

Fragment when you are assembling several parts and have the capacity. Finished plasmid when the construct is the deliverable, when time matters, or when the sequence verification is worth outsourcing along with the build.

**What should a plasmid preparation service return?**

Plasmid at a stated concentration and purity, with a supercoiled fraction, an endotoxin figure where it matters, an identity check and the strain it was grown in. A tube with a concentration alone is not enough for anything downstream.

**Are synthetic guides better than expressed ones?**

They are faster, transient and avoid integration, which suits primary cells and delivery as a complex. Expressed guides suit stable systems and library work. The choice follows the experiment rather than a general ranking.

**Why do suppliers ask about my institution?**

Because sequence screening and customer verification are a routine part of ordering synthetic DNA. Providing the details promptly is the fastest route through it, and orders from a new account should be placed with time in hand.

## Sanger sequencing

Short read sequencing did not make single read chemistry obsolete; it made it a specialist tool for a narrow and very common job. Confirming one region in a handful of samples, quickly, with a trace a person can inspect, is still done best the old way. Choosing between them is a question about sample count and read length, not about which is more modern.

1. **Count the samples and the regions**: A few regions across a few samples is single read territory: cheap per reaction, fast, and the trace is inspectable. Many regions or many samples crosses over to a short read platform quickly, and the crossover is lower than most people assume.
2. **Ask whether you need a consensus or the evidence**: A service returning only a called sequence has thrown away the information that lets you judge it. Ask for the trace files as well, because ambiguous positions, mixed peaks and poor early bases are visible there and nowhere else.
3. **Design the primers for the read, not for amplification**: Read quality is poor for the first stretch after the primer and degrades at the far end. Place primers so the region of interest sits in the reliable middle, and sequence both strands where a single base matters.
4. **Sequence whole constructs rather than junctions**: For plasmids, whole construct sequencing now costs little and catches rearrangements and unintended changes that junction reads miss. Reserve single reads for confirming a known edit at a known position.
5. **Record the read with the sample, not in an inbox**: Traces and consensus sequences belong in the construct or sample record under an identifier. Sequencing results filed as email attachments are results that will be paid for twice.

**Is short read sequencing more accurate?**

Per base in a deep dataset, yes, because coverage provides consensus. For a single amplicon read by an experienced eye, single read chemistry gives a long, contiguous, inspectable trace that is entirely adequate and far faster.

**How long a read should I expect?**

Several hundred usable bases after the primer, with quality falling at both ends. Design around the reliable middle rather than the theoretical maximum, and cover long regions with overlapping reads.

**What causes a mixed peak?**

A genuine heterozygous position, a mixed template such as an unpurified colony pool, or contamination from another amplicon. Distinguishing them requires the trace and often a re read from a clean template.

**Should plasmids still be verified this way?**

For confirming a known point change, yes. For accepting a newly built construct, whole plasmid sequencing is now the better economy because it catches the failures that a targeted read cannot see.

## Gel documentation

A gel imaging system is bought as a camera in a box and fails as a filter problem. Safer nucleic acid stains need different excitation and emission from the older intercalating dyes, and a system supplied with the wrong filter set will image your gels dimly and expensively. This page is about specifying the optics around the chemistry you actually use.

1. **Start from the stains in use, including future ones**: List every stain the laboratory uses for nucleic acids and proteins, and confirm the excitation source and emission filter for each. A system without the right pair for your stain is a system that will be worked around.
2. **Prefer epi illumination over a transilluminator where possible**: Blue or epi illumination avoids the damage and the safety issues of shortwave ultraviolet and suits the safer stains. Where an older stain must be imaged, confirm the system still supports it rather than assuming.
3. **Check saturation reporting for quantitative work**: If band intensity will be compared, the software must indicate saturated pixels and allow a raw export. Documentation only systems frequently apply automatic adjustments that make quantitation impossible.
4. **Decide whether blots belong in the same box**: Chemiluminescent blot imaging needs a cooled sensor and long exposures; gel documentation does not. A combined instrument is economical if both are needed, and an unnecessary expense if only gels are imaged.
5. **Automate the blot handling if throughput justifies it**: Incubation boxes and automated blot processors remove the uneven washing and drying that cause patchy membranes. For laboratories running blots continuously they improve consistency more than a better imager would.
6. **Confirm export and record keeping**: Raw images, acquisition settings and the analysis regions must be exportable and reviewable. An image that exists only inside a proprietary viewer is an image nobody can re examine.

**Why does my gel look faint on a new DNA gel imaging system?**

Almost always a filter mismatch with the stain. Confirm the excitation wavelength and the emission filter recommended for that specific dye, because manufacturers optimize their systems around their own stains.

**Do fluorescence imaging systems still need ultraviolet illumination?**

Only for older intercalating stains. Blue light illumination with the corresponding filters is safer for both users and DNA and is now adequate for most workflows, but it will not image every legacy stain.

**Can a fluorescent imaging system quantify bands?**

Only if it reports saturation and exports unadjusted images. Many do not, because they are built to produce a good looking picture, and automatic contrast adjustment is exactly what destroys quantitation.

**When does a capillary electrophoresis instrument replace a gel?**

When throughput, small sample volumes and consistent sizing matter more than seeing the separation. It removes gel pouring and human variability, and it costs considerably more per sample.

## Lab automation software

Automation projects rarely fail because a robot cannot move a plate. They fail because the scheduler cannot talk to one instrument, because nobody but the vendor can edit a method, or because the data lands somewhere the laboratory system cannot read. The software layer is where those failures live, and it is usually specified last.

1. **List every instrument and confirm a driver exists**: Ask for named, supported drivers for each instrument you intend to integrate, and ask who maintains them when the instrument's own software updates. An instrument without a maintained driver is an instrument outside the automation.
2. **Establish who writes methods after handover**: If method editing requires the vendor, every subsequent change is a change order with a lead time. Look for an editor a scientist can be trained on, and make training part of the purchase rather than an option.
3. **Design the data path before the workflow**: Decide where results land, in what format, and how they reach the laboratory system. Automation that produces files in a folder nobody parses has moved the manual work rather than removed it.
4. **Separate scheduling from execution**: A scheduler that optimises across instruments is a different product from the control software that drives one. Understand which layer you are buying, because a scheduler without device control still needs the drivers underneath it.
5. **Add recipe and batch record control where the process is regulated**: Manufacturing automation must hold recipes under version control and produce a batch record that ties parameters, deviations and results to a lot. Research schedulers do not do this, and retrofitting it is not realistic.
6. **Fix sample identity at the physical layer**: Automated identity depends on labels that machines can read reliably in the conditions they will meet, including cold and solvent exposure. A labeller and a barcode standard are unglamorous and prevent the failure mode that ruins automated runs.

**What is the commonest cause of a stalled automation project?**

An instrument that will not integrate, usually because its vendor provides no supported driver or restricts remote control. Confirm integration for every device before committing, with a demonstration rather than an assurance.

**Should the scheduler come from the robotic lab automation supplier?**

Not necessarily. Independent scheduling layers integrate across brands and reduce lock in; a supplier's own layer is usually simpler where the whole cell is from one maker. The deciding question is what else you will connect later.

**Is validation required for research automation?**

Not as such, but if the output supports a regulated decision, the electronic record expectations apply to the automation software as much as to the laboratory system. Deciding this early avoids an expensive retrofit.

**How much training does a laboratory need?**

At least two people able to write and debug methods, not one. Automation dependent on a single individual stops working when that person changes jobs, which is the quietest and most common failure in this category.

**Does clinical lab automation need different software from research?**

It needs the same scheduler and a great deal more record keeping: a result that leaves the laboratory as a report has to carry who approved it, on which instrument and against which calibration, and the software has to make an audit trail that cannot be switched off. The scheduling problem is the easier half in both settings.

## Target enrichment

Enrichment decides what the sequencer sees, which makes it the most consequential choice in a targeted experiment and the one most often left to a kit default. Capture and amplicon approaches differ in uniformity, in how much input they need and in how they behave on degraded material, and depletion and selection answer different questions about RNA entirely.

1. **Decide from input quality and quantity**: Amplicon approaches tolerate low and degraded input and give deep coverage of small regions. Capture needs more and better material and gives more uniform coverage across larger target sets. The material you actually have usually settles this.
2. **Judge on uniformity, not on percentage on target**: A high on target fraction with poor uniformity still leaves regions uncovered, and those regions are where the answer will be missing. Ask for coverage uniformity metrics from real samples rather than the headline enrichment figure.
3. **Choose RNA preparation by what you want to see**: Poly A selection captures mature messenger transcripts and discards everything else. Ribosomal depletion keeps non coding and unpolyadenylated species and costs more. For degraded material, depletion is usually the only workable route.
4. **Treat small RNA as a separate chemistry**: Short species need ligation based preparation, careful size selection and their own quality control. Running them through a standard preparation loses them, and normalising them against messenger references is not valid.
5. **Include the analysis in the design**: Duplicate handling, molecular identifiers and the variant calling approach all interact with the enrichment method. Agree the pipeline before samples are prepared, because some analysis choices require design features added at the library stage.

**Capture or amplicon?**

Amplicon for small panels, low input and fast turnaround; capture for larger target sets and better uniformity. Degraded material pushes toward amplicon, and a need to detect structural variation pushes toward capture or untargeted sequencing.

**When does enrichment stop being worth it?**

When the target set grows large enough that untargeted sequencing at adequate depth costs about the same. That crossover moves as sequencing gets cheaper and is worth recalculating rather than assuming.

**Poly A selection or ribosomal depletion?**

Selection for intact material where mature transcripts are the question, depletion for degraded material or where non coding species matter. They are not interchangeable and comparing data across them is a confound.

**Do molecular identifiers matter?**

For detecting low frequency variants, yes, because they distinguish true events from amplification artefacts. For straightforward genotyping they add cost and complexity for little benefit.

## Antibody discovery

A display campaign is a selection acting on whatever the library contains. If the library is small, biased or poorly constructed, panning enriches the best of a bad set and does so convincingly. Most disappointing campaigns are library problems or antigen problems, discovered after several rounds of selection have produced binders to the wrong thing.

1. **Judge the library on construction, not on a diversity number**: A stated diversity is an upper bound. What matters is the fraction of clones in frame and expressing, the absence of dominant sequences, and how the library was amplified. Ask for the quality control data behind the number.
2. **Make the antigen the way the target actually exists**: A correctly folded, correctly modified antigen presented in a relevant format is what determines whether the binders you select recognize the real target. Denatured or truncated antigen selects binders to something that does not exist in vivo.
3. **Design counter selection into every round**: Selection enriches binders to the plate, the tag, the linker and the carrier as readily as to the antigen. Counter selection against those, and against a closely related protein where specificity matters, is the difference between hits and artefacts.
4. **Screen at the format you intend to use**: A fragment that binds well on phage may lose affinity or express poorly when reformatted. Reformatting early, even for a subset, avoids carrying leads that cannot become the molecule you need.
5. **Sequence the output rather than picking clones blindly**: Sequencing the enriched population reveals dominant families, convergence and whether selection actually worked. Picking colonies without it means rediscovering the same clone repeatedly and missing rare, better ones.
6. **Check freedom to operate on the library and the format**: Libraries, display systems and some formats carry commercial terms that follow the molecule. Establish them before a lead exists, because a lead you cannot license is a lead you cannot use.

**Display or immunisation?**

Display gives speed, control of the selection conditions and access to non immunogenic or toxic targets. Immunisation gives affinity matured binders and broader epitope coverage. Many programs use both, and the choice depends on target and timeline rather than on principle.

**Why did panning give binders that do not work in my assay?**

Usually because the antigen presented during selection differed from the antigen in the assay, or because counter selection was inadequate. Selection is extremely good at finding whatever was actually there to bind.

**How many rounds of selection?**

Enough to enrich and not so many that a fast growing clone dominates. Monitoring enrichment between rounds and sequencing the population is more informative than a fixed number.

**What should a discovery service deliver?**

Sequences of the enriched families, the raw selection data, characterised binders in the intended format, and clear statements about library provenance and license terms. A handful of clones with affinity values is not a campaign record.

## Nucleic acid extraction

Extraction kits are chosen by species and sample type and should be chosen by what happens next. A preparation that is perfect for amplification can be useless for long read sequencing, because the property that matters there is fragment length and nothing in a standard quality check reports it. This page maps kits onto downstream requirements.

1. **Start from the downstream method's real requirement**: Amplification wants purity and freedom from inhibitors. Short read library preparation wants a reliable mass. Long read work wants intact high molecular weight material, which most standard kits actively destroy through column binding and vortexing.
2. **Match lysis to the sample, not to the catalog**: Bacterial cell walls, plant material and tough tissue need mechanical or enzymatic disruption that a mammalian protocol omits. Under lysis shows up as low yield; over aggressive lysis shows up as sheared material.
3. **Control the collection step**: Anticoagulant choice matters: some carry through and inhibit amplification. Storage temperature and time before extraction change both yield and integrity. Fix the collection protocol and treat it as part of the method.
4. **Check integrity as well as purity**: Absorbance ratios report contamination, not integrity, and a fluorescent assay reports mass, not length. Where fragment length matters, run a sizing method, because a preparation can score perfectly on the usual checks and be entirely sheared.
5. **Test for inhibition rather than assuming purity**: An amplification of a spiked control at two dilutions detects inhibitors that no spectrophotometer will see. It is the fastest way to distinguish a bad preparation from a bad assay.
6. **Standardize so preparations are comparable**: One kit, one protocol and one elution volume across a study. Mixed extraction methods introduce differences in yield, purity and length that propagate into every downstream measurement.

**Why does my DNA fail long read sequencing?**

Fragment length, almost always. Column based kits and vigorous mixing shear material below the length those platforms need, and no standard purity check reveals it. A dedicated high molecular weight protocol and gentle handling are required.

**What do absorbance ratios actually tell me?**

That protein or organic contamination is present or absent, approximately. They say nothing about concentration accuracy for a mixed sample, nothing about fragment length and nothing about inhibitors, all of which matter more.

**Magnetic beads or columns?**

Beads automate cleanly, handle variable volumes and are gentler on long fragments; columns are simple and cheap for small numbers. For automated or high throughput work beads are usually the better fit.

**Should extraction be standardised across a study?**

Yes. Different chemistries recover different fractions of the material and carry different residues, and mixing them within a study introduces a batch effect that looks like a result.

**Is a blood DNA extraction kit different from a tissue one?**

The lysis is easier and the inhibitor is specific: haem carries through and inhibits polymerases at concentrations an absorbance reading will not show. Kits for blood include a wash aimed at it, which is why using a tissue kit on blood produces clean looking material that amplifies badly.

**Does a DNA and RNA extraction kit compromise either?**

A little, and whether that matters depends on which one the experiment turns on. Splitting one lysate gives less of both and the RNA is the one that suffers, because the longer handling costs integrity. Where the sample is precious and both are needed it is the right tool; where there is enough material, two dedicated preparations are better.

## HPLC autosamplers

An autosampler is the part of a chromatography system that most often limits a method, and the part least often specified. Carryover between injections, precision at small volumes and the compatibility of the vial and septum with the sample together determine whether a quantitative method is capable at the low end. This page is about those details, and about buying used sensibly.

1. **Measure carryover, do not accept a specification**: Inject a high concentration standard followed by a blank and look at the blank. Carryover from needle, seat and loop is method and analyte dependent, and the number that matters is the one measured with your compound.
2. **Specify the injection volume range you will use**: Precision degrades at the bottom of an autosampler's range, exactly where trace methods operate. Choose a module whose comfortable range brackets your injection volume rather than one that merely permits it.
3. **Treat vials, caps and septa as method parameters**: Septum material can leach and can core; glass type affects adsorption of basic compounds; volume and insert geometry affect how low the needle can draw. Changing vial supplier without checking is an unrecorded method change.
4. **Understand what a method transfer really requires**: Scaling a method between particle sizes and column dimensions is arithmetic; making it work is not. System dwell volume, extra column volume and detector settings all change, and the transferred method needs revalidating rather than recalculating.
5. **Inspect a used instrument on the parts that wear**: Pump seals, check valves, injector rotor seal, needle and detector lamp hours. Ask for a recent performance qualification and run a real method during inspection, because a system that passes a pressure test can still fail on precision.
6. **Confirm software support before buying anything second hand**: An instrument whose control software will not run on a supported operating system is an instrument with a short remaining life, whatever its mechanical condition. This is the most common trap in a used purchase.

**What causes carryover?**

Analyte adsorbing to the needle exterior, the seat, the rotor seal or the tubing, and inadequate wash between injections. Fixing it usually means changing the wash solvent composition and the wash routine rather than replacing hardware.

**Do HPLC vials and caps really matter?**

Yes, particularly for basic compounds and at low concentrations, where adsorption to glass and leaching from septa are measurable. Fix the vial specification in the method and verify when a supplier changes.

**Is a used system a false economy?**

Not necessarily, but the saving should be judged against seals, a lamp, a service visit and the risk of unsupported software. A used instrument from a serviced fleet with records is a reasonable buy; an unknown one is a project.

**When is the higher pressure platform worth it?**

When run time or resolution genuinely limits throughput or the separation. Where methods are established, adequate and validated, migrating them costs more than the time it saves.

**Do hplc vial caps and inserts change the injection volume?**

Inserts do, because the needle depth is set for the vial and an insert changes where the liquid sits: too little volume in a tall insert and the needle draws air on the last injection of a sequence. Caps change nothing about volume and everything about whether the volume is still there an hour later. Set the needle depth for the insert you use and keep one geometry per sequence.

## GC supplies

Most gas chromatography troubleshooting ends at the inlet. Tailing peaks, ghost peaks, shifting retention and lost sensitivity are usually a liner, a septum or a leak rather than the column that gets replaced first. Treating consumables as a scheduled routine rather than a reaction to failure is the cheapest reliability improvement available.

1. **Match the liner to the injection mode and the analyte**: Split, splitless and on column injections need different liner geometry, and active compounds need deactivated surfaces and sometimes no wool. A liner chosen by habit is the commonest cause of tailing and of discrimination between components.
2. **Change septa on a schedule, not on failure**: A septum degrades with every injection and eventually leaks and sheds. The leak shows as retention drift and oxidised phase; the shedding shows as ghost peaks. Both are avoided by a count based replacement rule.
3. **Choose ferrules for the temperature and the fitting**: Graphite, vespel and blends behave differently under temperature cycling, and a ferrule that relaxes creates a slow leak that is difficult to find. Match the material to the maximum oven temperature and retighten after the first cycles.
4. **Specify gas purity and protect it with traps**: Oxygen and moisture destroy stationary phase and shorten column life dramatically. Traps are inexpensive next to a column, and they need replacing on a schedule too, which is easily forgotten.
5. **Select the column from the separation, then the method**: Phase polarity determines selectivity, film thickness determines capacity and retention of volatiles, and internal diameter determines efficiency and sample capacity. Method parameters can only optimize what the column makes possible.
6. **Keep a consumables log against system suitability**: Record what was changed and when, alongside suitability results. When a method drifts, the log usually contains the answer, and it is the evidence an investigation needs.

**Why are my peaks tailing?**

Active sites, most often in the inlet liner, followed by a poorly cut column end or a badly seated ferrule. Replacing the liner with a properly deactivated one and re cutting the column installation resolves the majority of cases.

**How often should the liner be changed?**

By injection count and sample dirtiness rather than by calendar. Dirty matrices may need it weekly; clean standards can run far longer. Establish a count from experience and write it into the maintenance schedule.

**Does carrier gas purity really matter?**

Considerably. Trace oxygen degrades the stationary phase at temperature, permanently, and the damage shows as rising baseline and lost efficiency. Traps and a stated gas grade are among the cheapest protections available.

**When is the column actually the problem?**

When inlet maintenance, a fresh cut and a leak check have all failed to restore performance, or when efficiency has fallen progressively over months. Replacing the column first is expensive and usually unnecessary.

## Electron microscopy

The instrument price in an electron microscopy quotation is frequently the smaller half of the first five years. Vibration isolation, power conditioning, cooling, the service contract, the sample preparation equipment and a trained operator are all necessary and are quoted separately or not at all. This page is about the whole cost and about when not to own one.

1. **Survey the site before shortlisting**: Floor vibration, stray magnetic fields, acoustic noise and temperature stability all limit achievable resolution, and remediation is a building project. A site survey before selection prevents buying performance the room cannot deliver.
2. **Cost the sample preparation separately**: Coating, polishing, sectioning, critical point drying or cryo preparation as the samples require, each with its own equipment and consumables. Preparation frequently costs more in time than imaging does, and it determines image quality more than the microscope.
3. **Decide whether elemental analysis is in scope**: Adding spectroscopy changes the detector requirement, the chamber geometry, the software and the training, and it imposes its own standards and calibration. It is a genuine capability rather than an accessory.
4. **Budget the service contract as a running cost**: Contracts on these instruments are a substantial annual figure, and running without one is a gamble against a single expensive failure. Get contract pricing during selection, not after.
5. **Name the operator**: These instruments reward skill heavily and punish neglect. An instrument without a named, trained custodian degrades and produces poor images that get blamed on the samples.
6. **Compare honestly against a shared facility**: For occasional imaging, a university or commercial facility gives a maintained instrument, an expert operator and no capital cost. Ownership makes sense when usage is high, when samples cannot travel, or when turnaround is critical.

**What does the quotation usually exclude?**

Site preparation, vibration and field mitigation, uninterruptible power, cooling, sample preparation equipment, the service contract and training. Together these often approach or exceed the instrument line in the first years.

**Is a benchtop instrument enough, or do I need the scanning electron microscope price of a full system?**

For routine morphology and quality inspection at moderate magnification, frequently yes, and the siting requirements are far lighter. High resolution work, cryo applications and demanding elemental analysis still need a full instrument in a proper room.

**How much does sample preparation matter?**

More than the instrument specification in most cases. Charging, contamination and artefacts from poor preparation limit far more images than the microscope's resolution does.

**Should occasional users buy or book?**

Book. A shared facility provides a maintained instrument and an operator who uses it daily, which yields better images than an owned instrument used monthly by someone who has to relearn it each time.

## Lateral flow assays

A lateral flow strip that works on the bench is a long way from a product. The membrane is a variable natural material, the conjugate is a colloid whose stability depends on the drying buffer, and the housing tolerances change the flow. Most scale up failures are traced to a material lot rather than to the assay design, and they surface late.

1. **Fix the membrane and qualify its lots**: Capillary flow rate varies between membrane lots, and it moves line position, sensitivity and background. Choose one membrane, agree a flow rate specification with the supplier, and test each incoming lot before it enters production.
2. **Develop the conjugate for stability, not just for signal**: Colloidal conjugates aggregate, and the drying buffer, sugars and blockers determine whether they survive storage. A conjugate optimized only for immediate signal will lose sensitivity within weeks.
3. **Treat every material as a controlled component**: Sample pad, conjugate pad, absorbent pad, backing card, housing and desiccant each affect performance. Change control on all of them, with a stated part number and supplier, is what makes lots comparable.
4. **Decide visual against reader based interpretation**: A visual read supports a qualitative claim; a reader supports a semi quantitative one and imposes its own validation, calibration and software obligations. Decide early because it changes the development and the labeling.
5. **Design the stability study before manufacturing**: Real time and accelerated storage at defined conditions, with a defined test schedule and acceptance criteria, is what supports a shelf life claim. Retrospective stability is not a thing.
6. **Transfer with a demonstrated process, not a protocol**: A manufacturing partner should run the process at their own scale, on their own equipment, with their own operators, and show three consistent lots before the transfer is accepted.

**Why does sensitivity change between batches?**

Membrane lot variation and conjugate stability, in that order. Both are managed by specification and incoming testing rather than by adjusting the assay, and a laboratory that re optimises each batch is not yet ready to manufacture.

**How is shelf life established?**

By real time storage at the labeled condition, supported by accelerated data, against defined acceptance criteria set before the study. Accelerated data alone supports a provisional claim at best.

**Does a reader change the regulatory position?**

It changes what can be claimed and adds software and calibration obligations. A semi quantitative result carries more expectation about precision and traceability than a visual yes or no.

**What should a manufacturing partner demonstrate?**

Three consecutive conforming lots at scale on their equipment, incoming material control including membrane flow testing, a documented process with defined critical parameters, and their own stability data on lots they made.

## siRNA and knockdown

Silencing an transcript is easy; attributing a phenotype to that silencing is not. Every sequence has off target activity, delivery reagents have effects of their own, and the timing between knockdown and phenotype is rarely considered. The controls that make the result interpretable cost more than the reagents and are routinely omitted.

1. **Order several independent sequences**: Two or preferably three non overlapping sequences against the same target, used separately. A phenotype seen with one sequence and not the others is an off target effect, and only independent sequences can distinguish the two.
2. **Include the controls that actually control something**: A non targeting sequence matched for chemistry and concentration, a delivery only condition, and where possible a rescue with a resistant version of the target. The rescue is the strongest evidence available and the least often done.
3. **Choose the modality by compartment and duration**: Short interfering duplexes act in the cytoplasm and are transient. Antisense chemistry reaches nuclear species and can work without a delivery reagent in some settings. The target's location and the required duration decide which.
4. **Solve delivery before optimizing sequence**: In hard to transfect and primary cells, delivery efficiency dominates outcome. Establish delivery with a labeled control and a well characterised positive sequence before concluding anything about your target.
5. **Measure knockdown at the protein, and time it properly**: Transcript reduction is not protein reduction, and a long lived protein can persist well past the transcript. Measure the protein, and choose the phenotype timepoint from the protein curve rather than from convention.
6. **Use the lowest effective concentration**: Off target effects scale with concentration. Titrate to the lowest dose that gives adequate knockdown rather than using a default, and report the concentration with the result.

**How many sequences are enough?**

At least two independent ones, three if the phenotype is the main claim. Pools are convenient and can mask a single dominant off target effect, so where a conclusion rests on the result, individual sequences are more informative.

**Is a rescue experiment always necessary?**

For a claim that a phenotype is caused by loss of a specific target, it is the strongest evidence and increasingly expected. It is not always feasible, and where it is not, independent sequences and a dose response are the fallback.

**Why does knockdown work in a cell line and not in primary cells?**

Delivery, almost always. Primary cells are harder to transfect and more sensitive to the reagents, which is why electroporation, conjugated chemistries or viral expression are used there instead of lipid transfection.

**Transcript or protein for confirming knockdown?**

Protein, because that is what the phenotype depends on. Transcript measurement is a useful early check and can be misleading on its own when the protein is stable.

**Which is the best siRNA transfection reagent?**

There is no general answer: efficiency and toxicity vary by cell type. Test a small panel on your own cells with a viability readout, and move to electroporation or a viral hairpin if the cells resist chemical transfection.

## AAV production and analytics

An adeno associated vector batch is described by a titre, and that titre is the least standardized number in the field. Whether the capsids carry a genome, whether that genome is complete, and which assay produced the figure all change the dose by large factors. This page is about specifying a batch so that what arrives can actually be used.

1. **Choose the serotype from the target, then check manufacture**: Tropism and immune exposure differ between serotypes, and so does how easily each is produced and purified. A serotype chosen purely on published tropism can be a difficult manufacturing program, and that is worth knowing before the construct is fixed.
2. **Name every titre method in the contract**: Genome titre, capsid titre and infectious titre differ by large factors and each depends on its assay design and standard. A batch specification that says a number without a method has specified nothing.
3. **Set an empty and partial capsid limit**: Capsids carrying no genome or a truncated one contribute to the immune load and not to the effect. Separating them is a real unit operation with a real yield cost, and the limit belongs in the specification rather than in the report.
4. **Build the purification around what the dose requires**: Affinity capture concentrates the vector; it does not separate full from empty. Density or ion exchange based polishing is what does that, and omitting it is the commonest gap between a research batch and a usable one.
5. **Agree the release panel before manufacture**: Identity, genome titre, capsid content, purity, residual host protein and DNA, residual plasmid, replication competent vector where applicable, endotoxin and sterility. Agreeing it afterwards means testing retained samples for methods that were not ready.
6. **Plan the analytics with the same lead time as the batch**: Assays that are not qualified when the batch is released delay everything. Analytical development belongs on the same timeline as process development, not behind it.

**Why do two laboratories report different titres for one batch?**

Different assay designs, different standards and sometimes different primer positions within the genome. It is the norm rather than the exception, which is why the method belongs in the specification and an independent cross check is worth running once.

**Do empty capsids matter?**

Yes. They add to the total capsid load without contributing effect, which affects the immune response and the interpretation of dose. Regulators expect the fraction to be measured and controlled, not merely reported.

**Is affinity purification enough, or do I need full AAV purification?**

For research use often yes. For anything dosed, no, because affinity resins bind the capsid regardless of what is inside it. A polishing step that separates by density or charge is required to control the full fraction.

**What lead time should be assumed?**

Longer than the manufacturing slot suggests, because plasmid supply, analytical qualification and release testing all sit around it. Booking a slot without those in place produces an expensive wait.

## Buffers and solutions

Buffers are prepared by whoever is free, recorded nowhere, and then blamed for nothing when an assay drifts. In practice the buffer is a reagent with a composition, a pH, an age and a preparation record, and several of the most persistent problems at the bench are buffer problems. This page treats them as reagents.

1. **Write the composition down and follow it**: Every buffer used in a method should have a written composition, a preparation procedure, a pH adjusted at the temperature of use, and a stated shelf life. A buffer made from memory is a variable nobody is tracking.
2. **Choose the staining buffer for the cells and the markers**: Protein content reduces non specific binding and can also quench some fluorophores; calcium and magnesium affect adhesion and some binding events; azide preserves and kills. Pick each component for a reason and keep it constant across a study.
3. **Match blocking to the membrane and the target**: Milk based blockers contain phosphoproteins and defeat phosphospecific antibodies. Protein free blockers suit those cases. The blocker is part of the method and changing it is a method change.
4. **Treat stripping as a last resort**: Stripping removes antibody and some of the target with it, and a stripped membrane is not equivalent to a fresh one. Where a second target must be measured, a fluorescent multiplex or a duplicate membrane is the better answer.
5. **Filter and degas anything going into an instrument**: Mobile phases and instrument sheath fluids need filtration and, for chromatography, degassing. Particulates cause the pressure and baseline problems that are usually blamed on a column.
6. **Label with date, maker and lot**: A bottle without a preparation date and a name is a bottle nobody can investigate. Labeling costs seconds and settles the commonest question in a drift investigation.

**Does the staining buffer really change results?**

Yes. Protein content, divalent cations, chelators and preservatives all affect binding, viability and background. A panel optimized in one buffer can behave differently in another, which is why the buffer belongs in the protocol.

**Milk or a protein free flow cytometry blocking buffer?**

Protein free for phosphospecific detection, because milk contains phosphoproteins that compete. Milk is cheap and effective otherwise. The choice is dictated by the target, not by preference.

**Can buffers be made from concentrated stocks?**

Yes, and it is usually the more consistent route, provided the dilution is done accurately and the pH is checked after dilution rather than assumed from the concentrate.

**How long do prepared buffers last?**

It depends entirely on composition and storage, and the honest answer is to set a shelf life per buffer and label accordingly. Solutions containing protein or reducing agents have short lives; simple salt solutions last far longer if sterile.

**Do flow cytometry tubes and plates behave differently?**

They do, and the difference is surface area and handling rather than the plastic itself. A plate loses more cells to the walls per volume and mixes less well, so a protein containing buffer matters more there. A plate is still the right answer for a panel across many samples, because the pipetting error a person makes across sixty tubes is larger than the loss.

**How should the fluorochromes for flow cytometry be assigned to markers?**

Brightest conjugate to the scarcest marker, and the dim ones to the markers present in large amounts, which is the opposite of how panels are usually assembled from what is in the freezer. Then check the spillover into the channels that carry your scarce markers before running samples, because that is the number that decides whether a population resolves.

## Cell based assays

Every cell based toxicity assay measures a proxy, and the proxies disagree with each other in informative ways. Metabolic activity falls before cells die and can also fall in cells that recover. Membrane integrity marks a late event. Counting cells is closest to the question and is the most work. Choosing among them is the experiment design.

1. **State the question as an endpoint**: Is the question whether cells die, whether they stop dividing, or whether a pathway is perturbed? Those are three different endpoints and three different assays, and a single viability number cannot distinguish them.
2. **Know what each dye actually reports**: Tetrazolium reduction reports metabolic activity. Nucleic acid exclusion dyes report membrane integrity. Protease and ATP based assays report other proxies again. Compounds that affect metabolism directly will confound the first without killing anything.
3. **Run a time course before fixing a timepoint**: Toxicity develops, and a single endpoint chosen for convenience can sit before or after the effect. One time course experiment makes every subsequent single timepoint interpretable.
4. **Control for compound interference**: Colored and fluorescent compounds interfere with plate readouts directly. Include compound only wells without cells, and check for absorbance or fluorescence at the read wavelength before concluding anything.
5. **Decide whether imaging is worth its pipeline**: High content imaging gives per cell information and mechanistic detail, and it brings an analysis burden that has to be resourced. Buy it when the mechanism matters, not because it produces more data.
6. **Decide in house against a contract laboratory**: Where results must be signed under a quality system, a contract laboratory brings the system, the trained staff and the documentation. In house work suits screening and mechanism where speed and iteration matter more.

**Why do my MTS assay and other viability assays disagree?**

Because they measure different things. Metabolic assays respond before cell death and can be affected directly by the compound; membrane integrity assays report a late event. Disagreement is information rather than a fault, and running two is often the right answer.

**Is counting cells better?**

It is closer to the endpoint most people mean, and it distinguishes reduced proliferation from cell death, which single dye assays cannot. It costs more effort per plate, which is why it is often reserved for confirming a hit.

**When is high content imaging justified?**

When the mechanism of the effect matters, when subpopulations are important, or when several endpoints must be measured on the same cells. It is not justified simply because a plate reader seems limited.

**Do these assays predict in vivo toxicity?**

Partially and inconsistently, which is why they are used for ranking and mechanism rather than for prediction. Being explicit about that in the study report avoids the result being over interpreted later.

## Modification state antibodies

Antibodies that detect a post translational modification are a different class of reagent from those that detect a protein. The epitope exists only in a modified state, it is removed by enzymes that stay active in a lysate, and the signal is frequently a distribution rather than a band. Most failures in this area are lysis failures rather than antibody failures.

1. **Inhibit the removing enzymes at lysis**: Deubiquitinases, phosphatases and other erasers remain active in a cold lysate and strip the modification within minutes. The inhibitor has to be in the buffer before the cells meet it, not added afterwards.
2. **Choose the reagent by linkage and by state**: Some antibodies recognize the modifier regardless of context, some recognize a specific chain linkage, and some recognize the modified form of one target protein. These answer different questions and are frequently sold under similar names.
3. **Expect a distribution, not a band**: Poly modified proteins run as a ladder or a smear because chain length varies. A clean single band from a general reagent is more likely to be a cross reaction than a result.
4. **Enrich when the fraction is small**: Only a small proportion of a target is usually modified at any moment. Immunoprecipitation of the target followed by modification blotting, or capture with a binding domain reagent, is often the only way to see it.
5. **Control with a treatment that changes the modification**: An inhibitor or stimulus that raises or lowers the modification is the control that proves the signal is the modification. An untreated lane alone proves nothing about specificity.
6. **Record the clone and the lysis buffer together**: For this class of reagent the buffer is as much part of the method as the antibody. Keep them recorded as a pair, because a protocol that omits the inhibitor cocktail will not reproduce.

**Why is my ubiquitination antibody blot a smear?**

Because chain length and site occupancy vary across the population, which produces a distribution rather than a band. For poly modified targets that smear is the correct result and quantifying it means integrating the region rather than picking a band.

**Why did the signal disappear?**

Almost always active removing enzymes during lysis. The inhibitor must be present in the buffer before lysis, samples kept cold and processing kept short. Warm handling for a few minutes is enough to lose the signal.

**Do I need a linkage specific reagent?**

Only if the linkage is the question. For a general question about whether a target is modified, a pan reagent with a good enrichment step is usually the practical route, with a linkage specific reagent to follow up.

**Can I strip and reprobe for the total protein?**

It is common and it is risky, because stripping removes target as well as antibody. A duplicate membrane or fluorescent multiplexing gives a more defensible total protein comparison.

**Why does a partial degradation look complete on my blot?**

Usually because the antibody was saturated or close to its detection limit at the starting level. Establish the linear range with a lysate dilution series first; outside it the blot reports the assay rather than the biology.

**Which lysis buffer should a turnover experiment use?**

It depends on the question. Harsh conditions release a protein from its complex and gentle ones keep it bound but may not solubilise it, so measuring the protein and measuring its partners call for different buffers. Using the laboratory default for both gives two different kinds of wrong answer.

## Signalling pathway antibodies

A signalling antibody is bought to report a state, and the state usually corresponds to a specific fragment, conformation or location rather than to the protein as a whole. Receptors that signal by cleavage, transcription factors that move compartments and membrane proteins that resist extraction each defeat a reagent chosen on the protein name alone.

1. **Decide which species of the protein answers your question**: Full length receptor, cleaved fragment, phosphorylated form or a conformational state are different analytes. For cleavage activated pathways the fragment is the signal, and an antibody against the extracellular region will never report it.
2. **Match extraction to the compartment**: Membrane and nuclear proteins need extraction conditions that a standard lysis buffer does not provide, and a target that appears absent is frequently a target that stayed in the pellet. Fractionate when the compartment matters.
3. **Choose fixation for the epitope, not for the tissue**: Cross linking fixation masks many epitopes and requires retrieval; some epitopes survive only in alcohol fixation or in frozen sections. The application listed on the datasheet is a starting point rather than a guarantee.
4. **Demand validation evidence in your application**: A blot on an overexpression lysate does not validate immunohistochemistry. Look for knockout or knockdown validation, and where it is absent, run a reduction control yourself before building on the reagent.
5. **Record clone and lot as part of the method**: Two clones against the same target are different reagents and frequently give different distributions. A protocol that names only the target is a protocol that will not reproduce when the catalog changes.
6. **Include a biological control that moves the signal**: A treatment or genotype expected to change the target is worth more than any isotype control. It shows the reagent tracks the biology rather than something constant.

**Why does an antibody work on a blot and not on a section?**

Because fixation and epitope accessibility differ completely. Denatured, linearised protein on a membrane presents epitopes that a cross linked tissue section hides, and retrieval conditions have to be optimized for each reagent.

**How do I detect an activated receptor with a Notch antibody?**

By detecting the species that activation produces, usually a cleaved fragment or a phosphorylated form, with a reagent raised against the neoepitope. Detecting total receptor tells you about expression, not activity.

**Is knockout validation necessary?**

It is the strongest available evidence and increasingly expected. Where no knockout exists, a knockdown with a clear signal reduction, or a peptide competition, is an acceptable substitute if documented.

**Why does my nuclear factor show only cytoplasmic staining?**

Frequently an extraction problem rather than biology. Nuclear proteins need a buffer that opens the nucleus, and standard whole cell lysis leaves much of the target in the insoluble fraction.

## Guide RNA

A guide is a short sequence and almost everything that goes wrong with it happens before it is ordered. Designing against the wrong genome build, against a region that differs in your cell line, or in a format the delivery method cannot use accounts for most failed edits. This page is about the decisions that precede the order.

1. **Design against the sequence you actually have**: Cell lines carry variants, and a guide designed on a reference genome may sit across a polymorphism in your line. Sequence the target region in your own cells before ordering when the edit matters.
2. **Choose the format from the delivery route**: Chemically synthesised guides suit ribonucleoprotein delivery and give a short exposure window. Plasmid or lentiviral expression suits stable systems, difficult cells and library work, and it integrates. The delivery route decides the format, not the other way round.
3. **Design several guides per target**: Cutting efficiency varies widely and is not reliably predicted. Ordering three guides and testing them is faster and cheaper than optimizing delivery around one that was never going to work.
4. **Plan the cutting assay before the experiment**: A mismatch cleavage assay or amplicon sequencing of the target region tells you whether the guide worked, separately from whether the phenotype appeared. Without it, a null result is uninterpretable.
5. **Size a library screen honestly**: Library work sets a representation requirement per guide that determines culture scale, transduction volume, sorting time and sequencing depth. Work those numbers before committing, because a screen below representation is noise.
6. **Store synthetic guides properly**: Chemically modified guides are stable dried and much less so in solution. Aliquot on first dissolution, record the buffer and avoid repeated freezing and thawing.

**Synthetic guide or an sgRNA plasmid?**

Synthetic for transient editing, primary cells and lower off target exposure. Expressed for stable systems, for continuous selection pressure and for library screens. Both work; the delivery constraint usually decides.

**How many guides should be tested per target?**

At least three, screened for cutting before any phenotype work. Efficiency is poorly predicted by design tools, and testing is cheap compared with troubleshooting a delivery system that was never the problem.

**Does the guide need modification?**

Chemical modification at the ends improves stability and reduces innate immune sensing, which matters in primary cells. For robust cell lines with plasmid delivery it is unnecessary.

**How is cutting confirmed?**

Amplify the target region and sequence it deeply enough to see the indel spectrum, or run a mismatch cleavage assay for a quick estimate. Confirming at the locus is separate from confirming loss of the protein, and both are needed.

**Which controls does an editing experiment need?**

A non-targeting guide delivered and selected identically, a positive guide with a known phenotype, more than one guide against the target, and a rescue where the conclusion rests on the edit.

**Is a mismatch-cleavage assay enough to report efficiency?**

No. It indicates editing happened and quantifies poorly, missing small or homogeneous edits. Sequence the amplicon for the allele spectrum, and for a clone sequence both alleles and confirm the protein is absent.

**Reporter library or an arrayed CRISPR screen with an sgRNA library?**

A reporter library measures what a sequence does to transcription, out of context, at large scale. An editing screen changes sequence in place and keeps the context. They are complementary, and the first nominates what the second tests.

## shRNA knockdown

Expressed hairpins give what transient reagents cannot: silencing that lasts through an experiment measured in weeks. They buy that with integration, with a selection marker and with a population that can drift as cells with less silencing outgrow those with more. Those trade offs are the whole decision, and they are manageable if planned.

1. **Decide transient against stable from the timescale**: Anything answered within a few days is better done with a transient duplex, which avoids integration entirely. Experiments running for weeks, or requiring the knockdown to persist through differentiation or in vivo work, need an expressed hairpin.
2. **Use several independent hairpins**: Off target activity applies to hairpins as much as to duplexes. Two or three independent sequences, used separately, distinguish a target effect from a sequence effect, and a non targeting hairpin controls for the vector and the selection.
3. **Deliver by the route the cells accept**: Lentiviral delivery reaches primary, suspension and non dividing cells that resist transfection, and it brings containment and institutional approval requirements. Plan those before the vector arrives.
4. **Watch for drift under selection**: Cells with weaker silencing frequently grow better, so a population knocked down at the start can recover over passages. Re measure knockdown at the end of every experiment, not only at the beginning.
5. **Confirm at the protein and record the timing**: Transcript reduction is not protein reduction, particularly for stable proteins. Measure the protein, and know how long after selection the knockdown reaches its floor.
6. **Rescue where the claim depends on it**: Re expressing a hairpin resistant version of the target and restoring the phenotype is the strongest evidence available that the phenotype is caused by loss of that target.

**Why has knockdown from my shRNA vector weakened over time?**

Selection for cells with lower silencing, silencing of the promoter driving the hairpin, or loss of the integrated cassette. All three are common, which is why knockdown should be confirmed at the end of an experiment rather than assumed.

**Pooled or individual hairpins?**

Individual sequences for anything a conclusion rests on, because a pool can hide a dominant off target effect. Pools are reasonable for a first pass when the aim is only to see whether the target is worth pursuing.

**Is a stable line necessary?**

Only when the experiment outlasts a transient reagent. Where a few days suffice, transient delivery avoids integration, selection and drift entirely and is usually the better experiment.

**What controls does a knockdown line need?**

A non targeting hairpin line taken through the same selection, at least two independent targeting hairpins, protein level confirmation at the time of the experiment, and a rescue where feasible.

**Why use an inducible hairpin?**

Because the same cells become their own control: uninduced against induced, same clone and passage. It also lets knockdown start after a state is established. Measure the leakiness and control the inducer's own effects.

## Culture media

Medium is the most consequential consumable in a cell culture laboratory and the one most often chosen by inheritance. Undefined components vary between lots, coatings interact with formulations, and a medium change part way through a program creates a comparability problem that is expensive to answer. This page is about choosing once.

1. **Prefer defined formulations where the work allows**: A defined, serum free medium removes the largest source of lot to lot variation and makes results comparable over years. Where serum is unavoidable, reserve lots and test them before committing.
2. **Treat the coating as part of the medium system**: Formulations are developed with a particular attachment substrate, and swapping the coating changes behavior even when the medium is unchanged. Validate the pair together, not separately.
3. **Reserve lots for the length of a study**: Ask the supplier to hold a lot, and buy enough of it to complete the study. A medium lot change mid study is a variable that cannot be removed from the analysis afterwards.
4. **Plan the transition to a graded medium early**: Moving to a manufacturing grade formulation changes documentation, supply terms and sometimes composition. Making that change while cells are already committed to a program is far harder than making it at the start.
5. **Record the lot with every experiment**: A medium lot number in the experimental record is what makes a later drift investigable. Without it, a change in behavior has no candidate explanation to test.
6. **Confirm the phenotype, not just the growth**: Cells can grow well in a medium that quietly permits differentiation or drift. Check the markers that define the state on a schedule rather than assuming healthy morphology means an unchanged population.

**Is serum free always better?**

For consistency and for anything heading toward the clinic, yes. For some primary and specialized cells, serum containing formulations remain the only ones that work, in which case lot testing and reservation become essential.

**Why did cells change behaviour on a new medium lot?**

Undefined components vary, and coatings and growth factors degrade at different rates. This is the reason to reserve lots and to record lot numbers, because otherwise there is nothing to compare against.

**What makes a medium GMP grade?**

Manufacture under a quality system with controlled raw materials, documented processes, release testing and traceability, plus supply terms that support a regulated process. The formulation may be identical to the research version.

**Why did cells change behaviour on a new stem cell growth medium lot?**

On a schedule set by how quickly the population drifts, and always before a critical experiment. Marker checks are cheap next to a study built on cells that were no longer what they were assumed to be.

## General lab equipment

A bench fit out is a list of instruments and a much larger list of consequences: consumables, service contracts, calibration, waste routes and the space each item needs to be used safely. Laboratories that buy from a list end up with equipment that cannot be serviced, cannot be sited, or costs more to run than to buy.

1. **Start from the protocols, not from a category list**: Write down the procedures the bench will actually run, then derive the equipment. A list assembled by category always includes things nobody uses and omits the specific item a real protocol needs.
2. **Cost consumables and service before the instrument**: For most bench equipment the purchase price is the smaller number over five years. Ask for consumable costs at your expected usage and for service contract pricing during selection, not after.
3. **Check the service route in your region**: An instrument with no local engineer and long parts lead times is an instrument that will be out of use for weeks at a time. Ask where parts come from and what a typical repair takes.
4. **Site cold storage as a facilities project**: Very low temperature freezers reject substantial heat, draw significant power and need an alarm and escalation plan with a defined response. Placing one in an unprepared room causes failures in it and in everything nearby.
5. **Separate clean and dirty flows in the layout**: Media preparation, culture handling, decontamination and waste need a direction of travel that does not cross. This is a layout decision that is nearly free at the planning stage and very expensive later.
6. **Put everything in one asset record with an owner**: Serial number, location, service and calibration status, and a named owner. Equipment without an owner is equipment nobody maintains, and it fails at the least convenient moment.

**What is most often forgotten in a fit out?**

Waste routes, service access around instruments, power and cooling for cold storage, and the consumables budget. All four are cheap to plan and disruptive to retrofit.

**Should equipment be bought new?**

Not necessarily. Simple mechanical items are good used purchases; anything with fluidics, optics or a controller needs service history and a support route, and unsupported software is the usual hidden trap.

**How should very low temperature freezers be protected?**

With continuous monitoring, an alarm that reaches a person who can act, a documented escalation, and either backup power or a rehearsed transfer plan. Sample value in these units typically exceeds every other asset in the room.

**Does laboratory equipment management need to hold calibration?**

If the instrument affects results, yes. A register of serial numbers answers the finance question; adding calibration and service status answers the scientific one and is what an audit will ask about.

## GPC and SEC

Size exclusion separation produces a curve, and every number quoted from that curve depends on choices someone made: the calibration standards, where the baseline was drawn, where the integration limits sat and which detector was believed. Two laboratories can run the same sample and report different averages while both being internally correct.

1. **State the calibration basis every time**: Conventional calibration reports mass relative to the standards used, and a polymer that differs in shape from those standards will be reported wrongly. Naming the standard set is not optional context; it is part of the result.
2. **Add light scattering where absolute mass matters**: A scattering detector measures mass directly rather than by comparison, which removes the standard dependence. It requires knowing the refractive index increment for the sample and it repays the effort where the value must be defended.
3. **Choose the column set for the whole distribution**: A column set resolves a range, and material outside that range is reported at the limits rather than accurately. Include the tails when choosing, because the tails are usually where the interesting behavior is.
4. **Fix baseline and integration limits in the method**: Where the operator places the baseline and the start and end of integration moves the reported averages substantially. Define them in the method rather than leaving them to judgement per run.
5. **Control temperature and solvent as method parameters**: Retention depends on both, and a solvent change alters the separation and sometimes the sample. Treat them as fixed method parameters with tolerances rather than as laboratory conditions.
6. **Run a check standard in every sequence**: A well characterised sample run alongside the batch shows whether the system and the integration are behaving. It is the only routine evidence that today's numbers are comparable to last month's.

**Why do two laboratories report different molecular weights?**

Different calibration standards, different integration limits or different column sets, usually all three. Harmonising the method, not the instrument, is what makes results comparable.

**Is light scattering always better?**

It gives absolute mass and removes standard dependence, at the cost of more method development and the need for accurate refractive index increment values. For routine comparative work conventional calibration is often adequate if the basis is stated.

**How much does integration matter?**

Considerably, particularly for the averages that weight the tails. Two analysts integrating the same chromatogram differently can report meaningfully different values, which is why the limits belong in the method.

**What should a report contain?**

The calibration basis and standards, the column set, solvent, temperature and flow, the detector configuration, the integration limits, and the check standard result for the sequence. Averages alone are not interpretable.

## Growth factor reagents

Growth factors are measured and applied as if they were single molecules, and most of them are families of isoforms distributed between free, matrix bound and receptor bound pools. An immunoassay sees some of those pools and not others, and a recombinant preparation delivers an activity that is only loosely related to its stated mass.

1. **Establish which isoforms your reagent sees**: A gene producing several splice variants needs an antibody whose immunogen region is present in the variants you care about. A pan reagent and a variant specific reagent answer different questions and are sold under similar names.
2. **Understand which pool the assay measures**: Free ligand, total ligand after dissociation, and ligand in complex give very different numbers from the same sample. A kit that does not state which it measures cannot be compared with one that does.
3. **Standardize sample handling for the analyte**: Platelet release during collection, clotting and freeze thaw all change measured concentrations for several growth factors. Fix the tube type, the processing time and the aliquoting, and keep them constant across a study.
4. **Titrate every new lot of recombinant protein**: Specific activity varies between lots and between suppliers even at the same nominal mass, because folding, aggregation and carrier protein content differ. Run a dose response with each new lot rather than matching by mass.
5. **Choose carrier free or carrier containing deliberately**: Carrier protein stabilises dilute preparations and interferes with some downstream applications. Decide which you need and keep it consistent, because switching changes effective concentration.
6. **Record lot with every experiment**: For this class of reagent, lot is the variable most likely to explain an unexpected result. Recording it costs nothing and is the first thing an investigation needs.

**Why do two human VEGF ELISA kit results give different concentrations?**

Because they measure different pools and calibrate against different standards. Neither is necessarily wrong, and the practical response is to choose one kit for a study and report the kit and standard with the values.

**Does isoform specificity matter, and is a VEGFA antibody different from a VEGF antibody?**

Whenever the biology depends on which isoform is present, which for several growth factor families it does. Check the immunogen region against the isoform structure rather than trusting the gene name on the label.

**Why does a new lot of recombinant protein behave differently?**

Specific activity differs between lots because of folding, aggregation and carrier content. Titrating each lot against a dose response, rather than matching stated mass, is the only reliable approach.

**How should samples be collected for growth factor measurement?**

With a fixed tube type, a fixed time to processing and minimal freeze thaw, because several of these analytes are released or degraded during handling. The collection protocol is part of the method.

## ADC development

A conjugate is three molecules and the chemistry that joins them, and the joins are where the program succeeds or fails. Linker stability decides whether payload arrives at the target or in circulation; conjugation control decides whether a batch is one product or a mixture; and the payload turns an ordinary biologics facility into a containment problem.

1. **Choose the linker against the release mechanism you want**: Cleavable linkers release payload under conditions found inside cells; non cleavable linkers release only on antibody degradation. Stability in circulation and release at the target are the two ends of the same trade, and the choice shapes everything downstream.
2. **Specify the distribution, not just the average ratio**: An average drug to antibody ratio can be produced by many different mixtures, and the unconjugated and heavily conjugated species behave differently in the body. The specification has to cover the distribution.
3. **Decide stochastic against site specific conjugation early**: Conventional conjugation is simple and produces a heterogeneous product. Site specific approaches produce a defined species and require antibody engineering and often a new process. Deciding late means redoing the antibody.
4. **Set limits for free drug and related impurities**: Unconjugated payload is a potent impurity with its own limit and its own assay. It must be controlled at release and monitored on stability.
5. **Treat the payload as a containment problem**: Highly potent compounds impose occupational exposure limits, containment engineering, cleaning verification and waste handling. Not every partner with biologics capacity can take this work.
6. **Build the analytical package before the batch**: Ratio and distribution, free drug, aggregation, identity, potency and stability indicating methods. These take longer to develop than the conjugation itself and are the usual critical path.

**Why do antibody drug conjugate linkers matter more than the payload?**

Because payload potency is largely known while linker behavior decides where that potency is delivered. Premature release produces systemic toxicity; excessive stability produces no effect at the target. The therapeutic window is largely a linker property.

**What does drug to antibody ratio actually tell you?**

An average. Two batches with the same average can contain very different proportions of unconjugated and heavily loaded species, which differ in clearance and toxicity. The distribution is the meaningful specification.

**Is site specific conjugation worth it?**

It gives a homogeneous product that is easier to characterise and often has a better profile, at the cost of antibody engineering and a more involved process. For programs intending to go far, it is increasingly the default.

**What should a conjugation partner demonstrate?**

Containment appropriate to the payload with occupational monitoring, an established conjugation platform with distribution control, free drug and aggregation analytics in house, and cleaning verification between products.

## Biologics manufacture

Two manufacturing routes are described with one word. Microbial fermentation is fast, dense and limited by oxygen transfer and heat removal. Mammalian culture is slow, dilute and limited by nutrient strategy and product quality. The molecule decides which is possible, and that decision constrains the facility, the analytics and the timeline that follow.

1. **Let the molecule choose the host**: Proteins requiring human like glycosylation, complex disulphide bonding or secretion of large multimers need mammalian culture. Simpler proteins that fold without those features are faster and much cheaper microbially, and the decision is effectively irreversible.
2. **Size oxygen transfer and heat removal for microbial routes**: Dense cultures consume oxygen and generate heat at rates that limit scale before volume does. Those two utilities, not vessel size, are what determine whether a process transfers to a given plant.
3. **Decide batch, fed batch or continuous with the analytics in mind**: Continuous operation raises volumetric productivity and shortens residence time, and it changes what a batch is, how release works and what in process control has to detect. It is a control strategy decision as much as an engineering one.
4. **Build comparability into a biosimilar program from the start**: A biosimilar has to match a reference product across an extensive analytical panel, which means the process is developed toward a target rather than toward a specification. That inverts the usual development sequence.
5. **Develop the analytical package alongside the process**: Identity, purity, charge and size variants, glycan profile, potency and the impurity panel. Characterisation is what turns a product into a well characterised one, and it takes longer than the process work.
6. **Fix the process before it becomes expensive to change**: Culture mode, host, purification sequence and critical parameters should be settled before clinical material is made, because changing any of them afterwards triggers a comparability exercise.

**When is microbial expression not an option?**

When the product needs human like glycosylation, when it is large and multi domain, or when correct disulphide formation cannot be achieved. Those constraints rule out the cheaper route regardless of economics.

**What actually limits microbial scale up?**

Oxygen transfer and heat removal, and mixing time in large vessels. Volume is rarely the binding constraint, which is why a process that works in a small fermenter can fail in a plant with different utilities.

**Does continuous processing reduce cost?**

It can raise productivity per unit volume substantially and reduce facility footprint, while adding control complexity and changing how batches are defined and released. The saving is real and it is not automatic.

**What makes a biologic well characterised?**

An analytical package extensive enough that product quality attributes are understood and controlled, rather than inferred from process consistency. It is a description of the evidence, not of the molecule.

## Cell lines and authentication

The most used cell lines in biology are also the most misidentified and the most quietly contaminated. Suspension lines punished by overgrowth, transfection hosts drifting under continuous selection, and lines swapped years ago in another laboratory all produce data that looks fine. Authentication and screening are what make a line's results mean anything.

1. **Control density rather than schedule**: Suspension lines have a working density range, and cultures allowed to overgrow change phenotype and respond differently for several passages afterwards. Count and split by density rather than by day of the week.
2. **Authenticate identity on arrival and periodically**: Short tandem repeat profiling against a reference database is inexpensive and definitive. Run it when a line arrives, before a major study and when anything about the line's behavior surprises you.
3. **Screen for mycoplasma on a schedule**: Contamination is common, invisible and changes cell behavior substantially. Test new arrivals in quarantine and cultures in continuous use on a fixed interval, and keep the results with the line record.
4. **Set and enforce a passage limit**: Lines change with passage number, and a study running across a wide passage range has an uncontrolled variable in it. Bank early, work from the bank, and return to it rather than culturing indefinitely.
5. **Bank properly and split the storage**: A frozen bank from a low passage authenticated culture, split across separate storage locations, is what lets a laboratory recover from a contamination or a freezer failure without losing years of work.
6. **Record the line's provenance**: Where it came from, when, its authentication result, its passage at receipt and its screening history. A line with no provenance is a line whose results cannot be defended.

**How often should cell lines be authenticated?**

On receipt, before publication or a major study, and at least annually for lines in continuous culture. Funders and journals increasingly ask for the evidence, and generating it retrospectively is not possible.

**Why do transfection host lines drift?**

Continuous passage and selection favor subpopulations, and transfection efficiency is one of the properties that changes. Working from a low passage bank rather than a long running culture keeps the behavior stable.

**Is a mycoplasma detection kit really necessary if cultures look fine?**

Yes, because contamination is invisible without testing and alters proliferation, gene expression and drug response. Testing only when something looks wrong means the affected experiments have already been done.

**What passage range should be used?**

A range narrow enough that behavior is stable, established for each line rather than assumed. Recording passage number with every experiment is the minimum, and returning to the bank rather than extending a culture is the discipline.

## HPLC systems

A chromatography system is bought module by module and behaves as a single instrument. Dwell volume, tubing dimensions, detector cell design and the software all interact, and a system assembled from individually excellent parts can be worse at your method than a modest matched one. This page is about specifying the whole thing.

1. **Choose the pressure class from the columns you intend to run**: Small particle columns need higher pressure capability and tighter plumbing; conventional columns do not. Buying a high pressure system to run conventional methods adds cost and complexity, and buying the reverse limits the columns available to you.
2. **Ask for dwell volume and plan gradients around it**: The volume between where the gradient is mixed and where the sample meets it delays every gradient, and it differs substantially between systems. Method transfer between systems with different dwell volumes fails for this reason more than any other.
3. **Minimize extra column volume deliberately**: Tubing lengths and diameters, fittings and the detector cell all broaden peaks after the column has separated them. On efficient columns this dominates the observed performance, and it is a plumbing decision rather than an instrument one.
4. **Match the detector to the analyte and the method's life**: A single wavelength detector is cheap and adequate for a fixed validated method; a diode array gives spectra and peak purity information that method development needs. Fluorescence and mass detection each impose their own mobile phase constraints.
5. **Cost the software and the qualification**: Chromatography data system licenses, instrument control drivers and the installation and operational qualification visit are substantial and are usually quoted separately. Ask for them at the shortlist stage.
6. **Plan system suitability before the first sample**: Define the suitability criteria the system must meet for each method, and confirm the candidate system meets them on your own sample during evaluation rather than on the vendor's test mixture.

**Why does the same method behave differently on another system?**

Dwell volume and extra column volume, in that order. Both shift retention and change peak shape, and neither appears in the method description. Recording both for each system makes transfers predictable.

**Is a higher pressure system always better?**

No. It opens up small particle columns and faster methods, and it demands tighter plumbing, more careful sample filtration and often new method validation. If your methods are established and adequate, the migration cost outweighs the benefit.

**Which of the HPLC detectors should a general purpose system have?**

A diode array, because it serves both routine quantitation and method development and gives peak purity information. Specialist detectors are added for named applications rather than for completeness.

**What is usually missing from a quotation?**

The data system license, the qualification visit, the service contract, columns and the consumables for the first year. Together they frequently approach the hardware cost.

**Why did my peaks broaden suddenly?**

Check the column head connection before the column. A fitting made up with the wrong ferrule depth leaves a void that destroys efficiency and looks exactly like a failing column; fittings from different makers are not interchangeable.

**Is a used system a sensible purchase?**

Often, since the mechanics are durable. What ends an instrument is unsupported control software on an unsupported operating system. Confirm software support and license transfer, then budget a service and a performance qualification.

## HPLC detectors

Detectors are chosen last and constrain everything. A fixed wavelength absorbance detector is cheap, robust and blind to whether a peak is one compound or two. A diode array sees the spectrum and settles that question. Fluorescence reaches concentrations absorbance cannot, for the minority of analytes that fluoresce. The right answer depends on whether you develop methods or only run them.

1. **Decide whether you develop or only run methods**: A laboratory running fixed validated methods needs sensitivity and reliability at known wavelengths. A laboratory developing methods needs spectra, peak purity assessment and the ability to find the best wavelength after the run.
2. **Check the cell volume against your peak widths**: A detector cell larger than the peak volume mixes the separation back together. On efficient columns and fast methods this is the dominant source of apparent inefficiency, and it is a specification you can ask for.
3. **Set the time constant and data rate for the peaks**: Heavy filtering smooths noise and also smooths narrow peaks. Fast methods need a high data rate and a short time constant, and defaults chosen for slow methods will silently distort them.
4. **Add fluorescence only for analytes that justify it**: Where the analyte fluoresces naturally or can be derivatised sensibly, the sensitivity and selectivity gain is large. Where it does not, a fluorescence detector is an expensive module that will sit idle.
5. **Use peak purity as evidence, not as proof**: Spectral purity assessment detects co elution when the spectra differ, and it cannot detect a co eluting compound with an identical spectrum. Treat a pass as supporting evidence alongside a second, orthogonal method.
6. **Record the detector settings with the method**: Wavelength, bandwidth, reference wavelength, data rate and time constant all change the chromatogram. A method that omits them is a method that will not reproduce on another system.

**Is a diode array HPLC PDA always worth the premium?**

For method development and for anything where co elution is a risk, yes, because the spectral information answers questions that would otherwise need extra experiments. For a laboratory running a small set of validated methods, a fixed wavelength detector is adequate and more sensitive at that wavelength.

**Why did my peaks get broader after changing detector?**

Cell volume, connecting tubing or the time constant. Detectors differ in all three, and the effect is largest on narrow peaks from efficient columns. Check the plumbing and the data rate before blaming the column.

**Does peak purity prove a peak is one compound?**

No. It shows that no spectrally distinguishable second compound is present. Compounds with similar spectra are invisible to it, which is why orthogonal confirmation matters for anything critical.

**When is mass detection the answer instead?**

When identity is the question rather than quantity, when the analyte lacks a useful chromophore, or when sensitivity requirements exceed what absorbance can reach. It brings mobile phase restrictions and a maintenance burden with it.

**Which of the detectors in HPLC should I buy first?**

The one that sees your analyte at the level you need. Absorbance for anything with a chromophore, fluorescence where sensitivity matters and the compound allows it, refractive index or conductivity where there is no chromophore at all.

**Can I run a gradient with a refractive index detector?**

No. It responds to any difference between sample and mobile phase, so a changing composition produces a changing baseline. It needs isocratic conditions and a thermostatted, well-equilibrated system.

## HPLC column chemistries

Column choice determines whether a separation is possible; method parameters only optimize what the chemistry allows. Analytes that do not retain on reversed phase will not retain however the gradient is adjusted, and charged species, enantiomers and large molecules each need a mechanism of their own. This page maps the mechanisms onto the analytes they suit.

1. **Start from how the analyte can be retained**: Hydrophobic interaction, charge, size and specific selector binding are the mechanisms available. Identify which one your analyte offers, because a column using a mechanism the analyte does not engage will not retain it at any gradient.
2. **Use exchange chemistry for charged species**: Anion and cation exchange retain by charge, which suits organic acids, nucleic acids, charged variants of proteins and inorganic ions. The retention is manipulated by salt and by pH, which makes buffer choice part of the column choice.
3. **Consider particle architecture, not only size**: Superficially porous particles give efficiency close to smaller fully porous ones at much lower back pressure, which lets a conventional system run faster methods. It is frequently the cheapest performance upgrade available.
4. **Treat chiral separations as a screening exercise**: There is no way to predict which chiral selector resolves a given pair. Screening a set of columns and mobile phases is the established route, and vendors will loan a screening kit.
5. **Match size separation to the range and the buffer**: Size based columns separate over a defined range in a defined mobile phase, and material outside that range comes out at the limits. For proteins the buffer must also keep the sample folded and non interacting.
6. **Record dimensions, chemistry, lot and history**: Two columns with the same nominal chemistry from different manufacturers are different columns, and even the same part number varies between production lots. Log injections and pressure so a decline is visible before it ruins a batch.

**When should reversed phase be abandoned?**

When the analyte is too polar or too charged to retain, when enantiomers must be separated, or when size rather than chemistry is the discriminating property. Trying to force retention with mobile phase alone wastes weeks.

**Are core shell columns a drop in replacement?**

Usually a drop in improvement rather than an exact replacement, because selectivity can shift slightly. For validated methods any column change requires revalidation, which is what makes the upgrade a project rather than a purchase.

**Strong or weak anion exchange HPLC?**

Strong exchangers stay charged across the usable pH range and are the general purpose choice. Weak exchangers change charge with pH, which gives an extra handle on selectivity and requires tighter pH control.

**Why did a replacement column give different retention?**

Lot to lot variation in bonding density and end capping, or a genuinely different chemistry sold under a similar description. Buy from one manufacturer for a validated method and record the lot with the results.

## Light microscopes

Most of a microscope's performance and most of its price live in the objectives, and most listings describe the stand. An instrument with an excellent frame and mediocre optics will disappoint in ways that are hard to articulate and easy to avoid. This page is about what to specify and what to check, new or used.

1. **Specify the objectives, then the stand**: Magnification, numerical aperture, correction class and the cover glass thickness they assume determine what you will see. A stand can be upgraded; a set of objectives is the majority of the cost and the whole of the image quality.
2. **Match the configuration to where the sample sits**: Cells in a flask need an inverted stand with long working distance optics and phase contrast. Slides need an upright stand. Buying the wrong orientation is the most expensive and most common mistake in this category.
3. **Check illumination and its alignment**: Even, correctly aligned illumination matters more than most specifications. For a used instrument, confirm the lamp type is still available and that any alignment adjustments are present and functional.
4. **Treat measurement configurations as calibrated instruments**: Any microscope used to measure needs a calibration against a traceable graticule for each objective and camera combination, repeated on a schedule. Without it the numbers are indicative rather than defensible.
5. **Confirm regulated configurations by their specification**: Certain analyses require a specific optical configuration, accessories and an accredited procedure. Buying a microscope that resembles the configuration is not the same as buying one that meets it.
6. **Inspect a used instrument physically**: Look for fungus and haze inside objectives, check focus mechanism smoothness and backlash, confirm the stage moves without drift, and verify that any camera and software still run on a supported system.

**What matters most in image quality?**

The objectives, followed by illumination alignment and the condenser. Camera resolution matters far less than people expect, because the optics set what reaches the sensor in the first place.

**Is a used microscope a good buy?**

Often yes for the stand and stage, which are mechanical and durable. Objectives and electronics are the risk: fungus, delamination and unsupported cameras and software are the usual problems, and all three are checkable before purchase.

**Do I need phase contrast?**

For unstained live cells, yes, and it needs matched objectives and a condenser with the right annuli. It is much cheaper to specify at purchase than to retrofit piece by piece.

**How is a measuring microscope calibrated?**

Against a traceable stage graticule, for every objective and camera combination in use, with the result recorded and repeated on a schedule. A single factory calibration is not sufficient once anything in the light path changes.

## Contract testing

Accreditation is granted for specific methods, not for laboratories in general, and the difference decides whether a report will be accepted by whoever asked for it. A laboratory can be genuinely expert in a technique and unaccredited for the method you need, which is fine for development work and not fine for a submission.

1. **Read the accreditation scope, not the website**: Accreditation bodies publish the exact methods each laboratory is accredited for. Check that your method appears, by standard number, rather than relying on a general claim of accreditation.
2. **Name the method in the enquiry**: Different standards for the same property give different numbers, and a laboratory quoting against an unnamed method will choose the cheapest. State the standard and the acceptance criteria in the request.
3. **Ask what is subcontracted**: Many laboratories subcontract part of a panel. That is acceptable when declared and disclosed on the report, and it affects turnaround, chain of custody and who is accountable for the result.
4. **Establish sample requirements before shipping**: Quantity, conditioning, packaging and how the sample must be identified. Samples rejected on arrival or conditioned incorrectly are the commonest cause of a delayed report.
5. **Compare turnaround including reporting**: Test time and report time are different, and the second is frequently longer. Ask for the quoted turnaround to be to a signed report, and ask what expediting actually costs.
6. **Decide in house against outsourcing on volume and defensibility**: Routine high volume testing with a stable method suits in house equipment. Occasional testing, unusual methods and anything requiring accreditation is almost always cheaper and safer outsourced.

**Does accreditation mean every test is accredited?**

No. It applies to a defined scope of methods, and laboratories routinely perform unaccredited work alongside accredited work. The report should distinguish them, and you should check which yours is.

**Why do two laboratories report different values?**

Different standards, different specimen conditioning or different sample preparation, in that order. Naming the standard and the conditioning in the request removes most of the discrepancy.

**When does in house equipment pay?**

When the same test is run often enough that per sample outsourcing exceeds the cost of the instrument, calibration and the person. Single purpose instruments used occasionally are usually a poor investment, because calibration continues whether they are used or not.

**What should a test report contain?**

The method by its standard reference, the specimen description and conditioning, the equipment used, the result with its uncertainty, any deviations, and a signature. A number without those is not a report.

**Is an accredited laboratory accredited for my test?**

Not necessarily. Accreditation covers specific methods on specific materials, listed on a schedule. Read the schedule rather than the certificate, and check for any sector approval your customer requires.

**What should I not do before a corrosion testing laboratory investigation?**

Do not clean, cut or handle the fracture surface. The part as found is the evidence; photographs, service history and the specification come before any sectioning, or the mechanism cannot be established.

## Purification kits

Purification kits are bought by sample type and should be bought by what the downstream method needs. Binding chemistry is broadly similar across brands; lysis is not, and neither is what the column does to fragment length. A kit that gives an excellent yield can be exactly the wrong kit for the method that follows.

1. **Choose lysis for the sample, binding for the target**: Blood, tissue, bacteria and plant material each need different disruption, and that is where kits actually differ. The binding step is comparatively generic, which is why kits marketed for different samples share a column and differ in buffers.
2. **Check what the format does to fragment length**: Spin columns shear long fragments; magnetic beads and gentle handling preserve them. If long reads or intact molecules are downstream, the kit format is the decision rather than a detail.
3. **Match plasmid kit scale to the actual requirement**: Miniprep, midiprep and maxiprep differ in culture volume and column capacity, and overloading a column silently reduces both yield and purity. For milligram quantities a preparation service is frequently cheaper than the consumables and the labor.
4. **Verify recovery, not just concentration**: A concentration reading tells you what came out, not what fraction of the input was recovered. Running a known input through the kit once establishes recovery and makes future low yields interpretable.
5. **Test for inhibitor carryover**: Chaotropic salts and ethanol carry over and inhibit downstream enzymes. A spiked amplification control at two dilutions detects this immediately and no spectrophotometer will.
6. **Standardize one kit per sample type**: Different chemistries recover different fractions and leave different residues. Mixing kits inside a study introduces a batch effect that is easy to avoid and impossible to remove afterwards.

**Are kits from different suppliers interchangeable?**

For robust downstream methods, largely yes. For sensitive ones they are not, because recovery, fragment length and residual carryover differ. Validate a change rather than assuming equivalence.

**Columns or magnetic beads?**

Beads for automation, variable volumes and gentler handling of long fragments; columns for simplicity and small numbers. Beads have become the default wherever throughput or fragment length matters.

**When should plasmid preparation be outsourced instead of using plasmid kits?**

When milligram quantities are needed, when a defined endotoxin level is required, or when the labor is displacing more valuable work. Below that, kits are simple and cheap.

**What single check catches most problems?**

A spiked inhibition control in the downstream reaction. It detects carryover, which is the failure mode that purity ratios miss and that gets misdiagnosed as low template.

## Storage and lab furniture

Storage is planned as a volume and used as an index. A freezer full of boxes nobody can navigate is worse than a smaller one that is mapped, because the search costs time and every open door costs temperature. The equipment around it, benches, mixers, hotplates and digesters, has its own siting and ventilation demands that are cheap to plan and expensive to retrofit.

1. **Choose the rack and box system first**: Usable capacity is a function of the rack, box and tube format, not of the cabinet volume. Fix one system across the laboratory so that racks are interchangeable and a position can be described unambiguously.
2. **Make every position addressable**: Freezer, rack, box, position, recorded against the sample. Without an address, retrieval means searching with the door open, which warms everything and eventually loses samples.
3. **Specify benches for the chemistry and the load**: Surface material has to resist the solvents and reagents in use, and the structure has to carry the heaviest instrument planned for it. Both are decided at order and are not adjustable afterwards.
4. **Site heating and digestion equipment with ventilation**: Hotplates, digestion units and anything generating vapour need extraction and clearance from flammables. Placing them where there happens to be a socket is how incidents happen.
5. **Use indicators for what they actually show**: An autoclave indicator tape shows a load has been exposed to a process, not that its contents are sterile. Sterility assurance needs a biological or chemical indicator placed where penetration is hardest, on a defined schedule.
6. **Write the storage plan down with an owner**: Who allocates space, what happens when a rack is full, how long material is kept and who authorises disposal. Storage without those rules fills up and then overflows into whichever freezer has room.

**What limits usable freezer capacity?**

The rack and box system and the discipline around it. A standardized system with an address for every position gives far more usable capacity than a larger unit filled with mixed formats and unlabelled boxes.

**Does an autoclave tape indicator prove sterility?**

No. It shows the load was exposed to the process. Sterility assurance requires an indicator placed in the hardest to reach part of the load, run on a defined schedule and recorded.

**How should bench material be chosen?**

By the chemicals used on it and the load carried. Solvent resistance, heat resistance and structural capacity are the three properties, and the cheapest surface is frequently the wrong one for the work.

**What should a retention policy cover?**

How long each class of material is kept, who authorises disposal, and how disposal is recorded. Without one, storage grows until a freezer failure disposes of everything at once.

## ELISA formats

The format of a plate immunoassay is decided by the analyte, not by preference. Small molecules cannot be sandwiched because two binders will not fit. Analytes with only one good binder available force a competitive design. Particles and vesicles need a capture step that works on something bigger than a protein. Choosing the format first makes everything after it easier.

1. **Check whether two binders can fit**: A sandwich needs capture and detection reagents binding distinct, simultaneously accessible epitopes. Small analytes physically cannot accommodate this, which is why they are measured competitively.
2. **Use competition when only one binder exists**: A competitive design measures inhibition of a labeled analyte's binding, which gives an inverted curve and generally poorer precision. It is the correct choice when a sandwich is impossible, not a fallback.
3. **Weigh direct against indirect detection**: Direct detection with a labeled primary is faster and has fewer cross reactivity risks. Indirect detection with a labeled secondary amplifies signal and lets one secondary serve many assays. Sensitivity usually favors indirect.
4. **Design capture for particles separately**: Vesicles and particles are captured through surface markers and are far larger than the analytes a plate assay normally handles, which changes coating, blocking and washing. A protein assay protocol applied to them will not work.
5. **Consider a kinetic read where rate carries information**: Reading development over time rather than at a fixed endpoint extends dynamic range and detects wells that saturated. It requires a reader that can do it and a plan for how the rate is calculated.
6. **Control washing before blaming the assay**: Residual volume and inconsistent washing cause more plate assay variability than any reagent. Fix the washer and the wash protocol before optimizing anything else.

**Why can't small molecules be measured in a sandwich format?**

Because two antibodies cannot bind a small analyte simultaneously without interfering. Competitive formats exist precisely for this case, and their inverted curve and narrower range are inherent rather than a defect.

**Are kinetic assays worth the extra time?**

When the dynamic range is limiting or when wells saturate at different times, yes, because a rate is informative where an endpoint has clipped. For routine assays inside a validated range, an endpoint read is simpler and adequate.

**What does the reader actually contribute?**

Accurate absorbance or fluorescence at the right wavelength with good well to well uniformity. Beyond that, assay quality is set by the reagents, the plate and the washing rather than by the instrument.

**How is a vesicle assay different?**

Capture is on a surface marker of a particle rather than on an epitope of a soluble protein, and the analyte is heterogeneous in size and composition. Standardisation and interpretation are harder, and the controls have to reflect that.

**Why would I use a competitive format?**

Because the analyte is too small to bind two antibodies at once, as haptens, small molecules and short peptides are. The signal then runs inversely to concentration and the working range is usually narrower.

**How do I compare two kits?**

On the validation report: the standard's traceability, sensitivity and range in your matrix, spike recovery, dilution linearity and precision. A cheaper kit with clean recovery in your matrix beats an expensive one without it.

## Detection reagents

Detection reagents are treated as interchangeable and are the layer where most unexplained background comes from. A biotin and streptavidin system amplifies beautifully and amplifies whatever biotin is already in the sample. Conjugates vary in labeling ratio. And a kit that bundles the layer hides the choices that would explain a failure.

1. **Decide whether amplification is needed at all**: A directly conjugated secondary is simpler and gives lower background. The biotin system adds a step and considerable signal, which matters for low abundance targets and is unnecessary for abundant ones.
2. **Block endogenous biotin where the sample carries it**: Tissues and cells contain biotin, and a biotin based detection layer will report it as signal. A blocking step with free avidin and biotin, or a different detection chemistry, is the remedy.
3. **Match the conjugate to the substrate and the imager**: Enzyme conjugates need a substrate suited to the sensitivity range and an imager that can capture it without saturating. Fluorescent conjugates need matching filters. Mismatches here waste the sensitivity you paid for.
4. **Titrate every detection reagent**: Excess conjugate raises background faster than it raises signal. Titrating the detection layer on your own samples usually improves the ratio more than any change of primary reagent.
5. **Keep capture chemistry separate in your thinking**: Affinity resins and capture reagents pull the target out of a mixture; detection reagents report that it is there. They share vocabulary and solve different problems, and conflating them produces protocols that do neither well.
6. **Record lots for the whole layer**: Primary, secondary, conjugate and substrate lots together determine the signal. When a blot changes, the layer is where the change usually is, and only a lot record makes that investigable.

**Why does my blot have high background with a streptavidin HRP antibody system?**

Either endogenous biotin in the sample or too much conjugate. Block endogenous biotin, then titrate the conjugate down. Both are quick and between them they resolve most cases.

**Direct or amplified detection?**

Direct for abundant targets, because it is simpler with less background. Amplified for low abundance targets where the extra layer buys real sensitivity, accepting the extra step and the biotin caveat.

**Does substrate choice matter?**

Considerably. Substrates differ by orders of magnitude in sensitivity and in how quickly the signal decays, and using a high sensitivity substrate on an abundant target guarantees saturation.

**Should I buy a kit or assemble the layer?**

A kit is convenient and consistent for a standard workflow. Assembling the layer gives control over each component, which matters when troubleshooting or when a component has to be titrated for an unusual sample.

**Is a secondary antibody HRP conjugate better than a biotin and streptavidin layer?**

For a strong signal, the direct conjugate is simpler and quieter: one binding step fewer means one source of background fewer. The biotin layer earns its place when the target is scarce enough that the extra amplification is the difference between seeing it and not, and the price is that everything else in the sample is amplified too.

## Difficult target antibodies

Some targets defeat antibodies for structural reasons: dense glycosylation that hides the peptide, membrane anchoring that resists extraction, cilia and matrix localisation that survives only particular fixation, and secreted proteins present in every buffer. Choosing a reagent for these is a matter of matching the immunogen to the obstacle.

1. **Identify the obstacle before the reagent**: Glycan masking, membrane anchoring, low abundance, high background from an abundant relative and localisation that fixation destroys are different problems. Naming yours narrows the reagent list immediately.
2. **Read the immunogen against the structure**: For repeat containing and heavily modified proteins, where the immunogen sits determines whether the epitope is accessible in your preparation. A datasheet naming only the protein is not enough.
3. **Match extraction to where the target lives**: Membrane anchored and matrix associated proteins need extraction conditions a standard lysis buffer does not provide. A target that appears absent is frequently a target left in the pellet.
4. **Choose fixation for the epitope and the structure**: Some structures survive only in particular fixation, and cross linking masks epitopes that then need retrieval. This is where most immunostaining of difficult targets fails and where a pilot on a positive control tissue pays for itself.
5. **Use a tagged construct as the validation shortcut**: Where the biology allows, expressing a tagged version and detecting the tag gives a clean positive control and calibrates what a real signal looks like. It is the cheapest validation available.
6. **Demand reduction evidence**: Knockout, knockdown or a tissue known to lack the target is what distinguishes a validated reagent from a sold one. Where the supplier has none, generate it yourself once and keep the image with the reagent record.

**Why does an antibody fail on a glycosylated target?**

Because the glycan physically blocks the peptide epitope in the native protein, even though the immunogen worked. Reagents raised against accessible regions or against the glycan itself, and deglycosylation controls, are the routes around it.

**Is a tag antibody a substitute for a target antibody?**

Only for the tagged construct, and it says nothing about the endogenous protein. It is extremely useful as a positive control and as a calibration of what real signal looks like.

**How much validation is enough?**

Enough that someone else would believe the signal is the target: a reduction control, an appropriate positive and negative sample, and consistency between two independent reagents where the conclusion depends on it.

**What if no validated reagent exists?**

Say so in the write up and choose a design that does not depend on the antibody alone, for example a tagged construct, a genetic readout or mass spectrometry. An unvalidated reagent presented as validated is the worse option.

## Inflammasome antibodies

Innate immune pathways are activated by proteolysis and by assembly, not by transcription, so an antibody that reports abundance reports nothing about whether the pathway fired. The reagents that matter detect a cleaved product, a neoepitope or a redistribution, and the sample preparation has to preserve the state you are trying to read. This page is about choosing on that basis.

1. **Buy against the cleaved form, not the gene**: An enzyme that circulates as an inactive precursor is only informative once cleaved. Reagents raised against the neoepitope exposed by cleavage answer the activation question; reagents against the parent detect both forms and are useful only as a loading reference.
2. **Read assembly as redistribution, not as amount**: Adaptor and sensor proteins signal by forming specks and filaments. Their total abundance may not change at all, so imaging the distribution, or a native gel, tells you what a denaturing blot cannot.
3. **Preserve the state at lysis**: Proteolysis continues in a lysate and phosphorylation is stripped within minutes. Inhibitors have to be in the buffer before the cells meet it, samples kept cold and processing kept short, or the pathway will look inactive whatever it was doing.
4. **Measure the secreted product as well as the cell**: Mature cytokines leave the cell, so a lysate blot can miss the output entirely. Pair the cellular readout with a supernatant measurement, and be explicit about which form the assay detects.
5. **Control with a stimulus and an inhibitor**: A treatment that fires the pathway and one that blocks it bracket every experiment. Without them a band cannot be distinguished from a background, and a negative result cannot be distinguished from a failed stimulation.

**Why does my caspase blot show only the precursor?**

Either the pathway did not fire, or the cleaved product left with the supernatant, or the cleavage was reversed during handling. Blot the supernatant as well as the lysate and check the inhibitor cocktail was in the buffer before lysis.

**Can abundance report inflammasome activation, or do I need an NLRP3 antibody on assembly?**

No. Assembly is the activation event and it does not require new protein. Imaging speck formation, or a native separation, is what reports it; a denaturing blot of total protein does not.

**Which controls make these results credible?**

A known stimulus, a specific inhibitor, and where possible cells lacking the sensor. Reduction of signal in the knockout is the strongest evidence a reagent is reporting what its label claims.

## Apoptosis antibodies

No single antibody reports apoptosis. The survival proteins set a threshold, the sensitisers push against it, the caspases execute and the cleaved substrates record that execution happened. Measured alone, each is ambiguous; measured together, they describe a decision. This page is about assembling that set and reading it honestly.

1. **Measure the balance, not one side of it**: A survival protein rising means nothing without knowing what the sensitisers did. Blot at least one from each side on the same membrane, and report the ratio rather than the individual intensities.
2. **Use cleaved substrate as the execution readout**: A cleaved substrate is direct evidence that the machinery ran. Total substrate is a loading reference. Confusing the two is the most common misreading in this area, and reagents for both are sold under similar names.
3. **Separate the two initiating routes**: Receptor initiated and mitochondrial routes converge on the same executioners, so an executioner readout cannot tell you which fired. Blotting the two initiators, or blocking one, is what distinguishes them.
4. **Collect the cells that detached**: Dying cells lift off, and a lysate made from the adherent layer alone systematically under reports death. Collect the medium, spin it and add those cells back before lysis.
5. **Confirm with an orthogonal readout**: A surface stain, a fluorogenic activity assay or a viability count alongside the blot turns a suggestive band into a result. Two methods agreeing is what makes this class of experiment credible.

**Why is my caspase blot negative when the cells clearly died?**

Frequently because the dying cells detached and were washed away before lysis, or because the timepoint sat after the peak. Collect the supernatant cells and run a time course before concluding the pathway was not involved.

**Total or cleaved reagents, and where does a PARP1 antibody fit?**

Cleaved for the question of whether death occurred, total as the loading reference. Buying only the total reagent is the commonest way to end up with a blot that cannot answer the question asked of it.

**Is a survival protein level meaningful on its own?**

Rarely. It sets a threshold that only means something relative to the sensitisers present. Report both, and treat any single value as descriptive rather than mechanistic.

## Proliferation antibodies

A proliferation marker tells you a cell is not resting. It does not tell you how fast the population divides, and several of the most used markers persist well past the event they mark. Reading them as a rate, or treating two of them as interchangeable, produces conclusions the staining cannot support. This page is about what each one actually reports.

1. **Decide whether you need cycling or division**: Markers of the cycling state identify cells that are not quiescent. Measuring division requires incorporating a label over a defined window, which is a different experiment with a different reagent and a denaturation step.
2. **Match the marker to the phase you care about**: Licensing factors, phase specific cyclins and mitotic kinases each peak at different points. A marker chosen without regard to phase gives a number that mixes populations and cannot be compared across conditions that shift the distribution.
3. **Fix fixation and retrieval per reagent**: Nuclear proliferation markers are sensitive to fixation time and retrieval conditions, and a weak stain is far more often a retrieval problem than a reagent problem. Optimize on a tissue known to be proliferating.
4. **Score consistently and say how**: Percentage of positive nuclei, intensity thresholds and how partially stained nuclei are treated all change the number. Fix the scoring rule, apply it blind, and state it with the result.
5. **Include an arrested control**: A serum starved or confluent population gives the low end of the scale and shows the stain is reporting biology rather than background. It costs one well and settles most disputes about a threshold.

**Does a proliferation index measure growth rate?**

No. It measures the fraction of cells in cycle at one moment. A population with a long cycle can show a high index and divide slowly, which is why comparing indices between very different cell types is unsafe.

**Why do two proliferation markers, such as a Ki67 antibody and a p16 antibody, disagree?**

Because they mark different things: one may persist after the cycle stops, another may be degraded rapidly at a specific phase. Disagreement is usually informative rather than an error.

**When is nucleotide incorporation the better choice?**

When the question is how many cells divided during a defined window, rather than how many were in cycle when the sample was taken. It requires the label, the exposure window and a denaturation step the marker stains do not.

## DNA damage antibodies

Damage markers report the response, not the lesion. A phosphorylated histone marks where signalling spread around a break; a recombinase marks a route being attempted; an oxidised base marks chemistry that may never have become a break at all. Choosing among them is choosing which part of the story you want, and the readouts are foci rather than band intensities.

1. **Decide lesion, signal or route**: An oxidised base, a signalling mark spreading from a break, and a repair factor loading are three different observations. Naming which you need selects the reagent and also the readout, because only some of them are meaningful as foci counts.
2. **Count foci rather than measure intensity**: The signalling mark spreads over a large chromatin domain around each break, so the informative measurement is the number of discrete foci per nucleus. Whole nucleus intensity mixes that with background and with cycle stage.
3. **Control for replication**: Cells in S phase carry damage marks and repair factor foci as a normal part of replication. Without a cycle marker in the same image, an apparent increase in damage can simply be more cells replicating.
4. **Use a defined dose and a time course**: Foci appear within minutes and resolve over hours, and the resolution curve is the informative part. A single late timepoint reports repair capacity as if it were damage.
5. **Include a positive damage control on every slide**: A known dose of a damaging agent on the same slide calibrates the stain and shows the reagent worked. Damage experiments without one are difficult to interpret across days.

**Does a damage marker measure the number of breaks?**

Approximately, at early timepoints and low doses, when foci are resolvable. At high dose foci merge and the count saturates, and at late timepoints the count reflects unrepaired damage rather than initial damage.

**Why do untreated cells show foci with a RAD51 antibody or gamma H2AX antibody?**

Replication generates damage signals as a normal part of copying the genome. Staining a cycle marker alongside, and gating on non replicating cells, separates that from induced damage.

**Foci or western blot?**

Foci for spatial information and per cell distribution, which is what most damage questions need. A blot averages across the population and can miss a large response in a small subpopulation entirely.

## Chaperone and stress antibodies

Chaperones are abundant proteins whose interesting behavior is induction, relocalisation and client handling rather than presence. Several families contain a constitutive member and an inducible one that differ by a few residues, so a reagent that does not distinguish them will report a large constant signal and miss the response entirely.

1. **Name the family member, not the family**: Chaperone families contain closely related members with different regulation. Check that the reagent's immunogen distinguishes the inducible member if induction is the question, and be sceptical of a pan family reagent used to claim a stress response.
2. **Match the compartment**: Cytosolic, mitochondrial and endoplasmic reticulum chaperones are different proteins with different markers. A generic organelle reagent is usually a poorer choice than a named resident protein, because you can then say what it marks.
3. **Read the unfolded protein response as branches**: The response has separate arms with separate readouts, and a stress can activate one and not another. Measuring one branch and reporting the response as a whole is a common overstatement.
4. **Include a positive stress control**: A known inducer on the same blot establishes the dynamic range and shows the reagent detects induction. Without it a modest change cannot be placed on any scale.
5. **Watch loading controls under stress**: Common loading references change under stress, which produces spurious normalisation. Total protein staining is safer for this class of experiment than any single housekeeping protein.

**Why is my chaperone signal huge and unchanging?**

Almost certainly a reagent detecting the constitutive family member alongside the inducible one. Switch to a reagent whose immunogen distinguishes them, or accept that the blot cannot report induction.

**Which branch of the unfolded protein response should an ATF4 antibody or XBP1 antibody measure?**

Whichever the hypothesis concerns, and preferably more than one, because stresses activate them differentially. Reporting one branch as the whole response is the most frequent overstatement in this area.

**Are housekeeping loading controls safe here?**

Often not, because stress changes their expression too. Use total protein staining for normalisation in stress experiments, and show the total protein image.

## JAK STAT antibodies

Cytokine signalling runs on phosphorylation, which appears within minutes and is stripped within minutes of careless handling. A total protein reagent tells you the pathway component is present; only the phosphorylation specific reagent tells you it fired, and only if the sample reached denaturation fast enough to preserve it.

1. **Buy phospho and total as a pair**: The phosphorylated fraction only means something as a proportion of the total, and both have to come from the same sample. Buying only the phospho reagent produces a result nobody can normalise.
2. **Put phosphatase inhibitors in the buffer first**: Phosphorylation is removed in a cold lysate within minutes. The inhibitor has to be present before the cells meet the buffer, and the path from plate to denatured sample has to be short.
3. **Run a stimulation time course**: Signalling peaks and decays on a scale of minutes. A single timepoint chosen for convenience regularly lands after the peak and reports a negative from a positive experiment.
4. **Include a stimulated and an inhibited control**: A ligand that fires the pathway and an inhibitor that blocks it bracket the assay and show the reagent tracks the biology. This pair is worth more than any isotype control.
5. **Measure ligands where they go**: Secreted cytokines leave the cell, so a lysate is the wrong sample. Measure the supernatant or the circulation, and state which form of the ligand the assay detects.

**Why is my phospho signal absent?**

Handling, in the great majority of cases: no phosphatase inhibitor, warm buffer, or too long between the plate and the denaturation step. Fix the lysis protocol before changing reagents.

**Do I need the total reagent too?**

Yes. The phosphorylated fraction is interpretable only relative to how much protein is present, and treatments frequently change both. Reporting phospho alone is uninterpretable.

**Should cytokines be measured in lysate?**

No. Secreted mediators are measured in the supernatant or in circulation. A lysate reports what has been made and not yet released, which is a different and usually less useful quantity.

## NF-kB pathway antibodies

This pathway is activated by releasing a factor that was already there. Total protein does not change; what changes is where the factor is and whether its inhibitor has been destroyed. A whole cell lysate blot is therefore close to blind, and the reagents worth buying are the ones that let you see location and degradation.

1. **Fractionate or image, do not blot whole lysate**: Translocation is the activation event. Nuclear and cytoplasmic fractions on the same blot, or imaging of nuclear intensity per cell, are the two readouts that report it. A whole cell lysate averages the answer away.
2. **Use inhibitor degradation as the early readout**: The inhibitor is destroyed within minutes of stimulation and resynthesised afterwards. Its disappearance is a fast, unambiguous activation signal and needs a tight time course to catch.
3. **Validate the fractionation itself**: A nuclear fraction contaminated with cytoplasm invalidates the whole experiment. Blot a known nuclear and a known cytoplasmic marker on every fractionation, and show them in the figure.
4. **Expect oscillation and design around it**: The pathway oscillates rather than switching once, so a single late timepoint can catch a trough. Sample early and often, and report the time course rather than a representative lane.
5. **Control with a receptor level block**: Blocking at the receptor, or using cells lacking it, distinguishes pathway activation from a general stress response that moves the factor for other reasons.

**Why does the total factor level not change with an NFkB p65 antibody or RelA antibody?**

Because activation is release and relocalisation, not synthesis. The factor is present in the cytoplasm bound to its inhibitor before stimulation and in the nucleus after it, and a whole cell blot sees the same total either way.

**What is the fastest reliable readout?**

Disappearance of the inhibitor, which happens within minutes and is easy to see on a blot. It requires a time course fine enough to catch it before resynthesis.

**How do I show a fractionation was clean?**

Blot a nuclear marker and a cytoplasmic marker on the same membrane and include them in the figure. A translocation claim without those controls is not evidence.

## Developmental signalling antibodies

Developmental pathways activate by releasing, cleaving or relocating proteins that were already present, and several of their central molecules sit in two pools doing two jobs. A reagent that cannot tell those pools apart reports a large constant signal from the structural pool and misses the signalling pool entirely.

1. **Separate the structural pool from the signalling pool**: For proteins that also sit at cell junctions, most of the signal in a lysate is structural. Reagents raised against the non phosphorylated signalling form, or fractionation and imaging, are what reveal the pool that matters.
2. **Detect cleavage products, not receptors**: Where a pathway activates by proteolysis, the intracellular fragment is the signal. An antibody against the extracellular region of the receptor will never report activation.
3. **Treat translocation pathways as imaging problems**: Factors that move between cytoplasm and nucleus need fractionation with validated markers, or per cell imaging. A whole cell blot averages the event away.
4. **Measure secreted ligands where they act**: Ligands in these pathways are lipid modified, poorly soluble and act locally. Lysate measurements are unreliable and conditioned medium is often the only informative sample.
5. **Anchor with a pathway agonist and antagonist**: A known activator and a known inhibitor bracket the assay. Given how much of the signal in these pathways is constitutive, that bracket is what makes a modest change believable.

**Why is my signal high in unstimulated cells?**

Because most of the protein is in a structural pool at junctions or in a stable complex. Use a reagent specific to the active form, or a subcellular readout, rather than trying to interpret total lysate.

**How do I detect activation of a cleavage pathway, for example with an NICD antibody?**

With a reagent raised against the neoepitope created by cleavage, and by looking in the nucleus rather than at the membrane. Total receptor tells you about expression only.

**Can these ligands be measured in a lysate?**

Poorly. They are secreted, lipid modified and act over short distances, so conditioned medium or a reporter assay is usually a better measurement than any blot.

## Reporter and tag antibodies

Antibodies against introduced proteins are the cleanest reagents in a laboratory, because a cell line without the construct is a perfect negative control. That advantage is routinely wasted: the negative line is not run, cross reactivity within a fluorescent protein family is assumed away, and a tag is detected without checking the fusion is intact.

1. **Run the untransduced line every time**: A cell line lacking the construct is the ideal negative control and it costs one lane. Any signal there is background, and knowing its size makes every subsequent interpretation quantitative.
2. **Check cross reactivity across the family**: Fluorescent proteins share substantial sequence, and a reagent raised against one frequently detects several. If two colors are used in the same experiment, confirm the reagents distinguish them before relying on either.
3. **Confirm the fusion is intact**: A tag antibody reports the tag, and a cleaved or degraded fusion still carries it. Run the tag reagent and a reagent against the partner and check the bands agree in size before treating a tag signal as the protein.
4. **Choose the tag for the application**: Small peptide tags are less likely to perturb function and are harder to detect at low abundance. Larger fusion tags detect easily and are more likely to interfere. Decide against the experiment rather than by habit.
5. **Keep detection chemistry separate in your thinking**: Reagents against labels such as biotin, digoxigenin or a fluorophore are part of a detection layer. Treating them as target reagents leads to protocols where nobody can say what the signal reports.

**Do I still need controls for a tag antibody?**

Yes, and they are unusually easy: an untransduced line, and where possible a construct with the tag in another position. The convenience of the reagent is not a reason to skip the evidence.

**Will an EGFP antibody against one fluorescent protein detect another?**

Frequently, because the family shares sequence. Test it explicitly if two are used together, because a cross reacting reagent will quietly merge the two populations.

**Why does my tag blot show the wrong size?**

Degradation of the fusion, an unintended start site, or cleavage at a linker. Blot the partner as well and compare, because a tag signal at an unexpected size is information rather than noise.

## Secondary antibodies

A secondary reagent is defined by three things: the species and chain of the primary it binds, the species it was raised in, and what it has been adsorbed against. Get any of those wrong and the result is background that looks like signal, or two channels that quietly report the same thing. This page is about reading the name properly.

1. **Match the chain, not just the species**: A reagent against whole immunoglobulin binds both heavy and light chains and therefore cross reacts with other classes sharing light chains. A reagent against a specific heavy chain or a specific fragment is narrower and is what multiplexing requires.
2. **Choose the host to avoid the sample**: The secondary must be raised in a species unrelated to the sample and to any other reagent in the experiment. Staining mouse tissue with a mouse primary is the classic trap, and it needs a dedicated reagent system rather than an ordinary secondary.
3. **Insist on cross adsorption for multiplexing**: Cross adsorbed reagents have been passed over immunoglobulin from other species to remove antibodies that would bind them. In a two color or two species experiment this is not an upgrade, it is a requirement.
4. **Match the conjugate to the readout**: Enzyme conjugates for chemiluminescent and chromogenic detection, fluorophores for multiplexed quantitation, biotin where amplification is needed and endogenous biotin has been blocked. The conjugate decides the imaging as much as the reagent does.
5. **Titrate the secondary on your own samples**: Excess secondary raises background faster than signal. Titration usually improves a blot or a section more than any change of primary, and it is a single plate or membrane of work.

**What does cross adsorbed actually mean?**

The reagent has been passed over immunoglobulin from other species so that antibodies binding those species are removed. It matters whenever more than one species is present in the experiment, including the tissue itself.

**Can I stain mouse tissue with a mouse anti human antibody?**

Only with a system designed for it, because an ordinary anti mouse secondary will bind the endogenous immunoglobulin throughout the tissue. Dedicated kits and directly conjugated primaries are the two workable routes.

**HRP conjugated antibodies or a fluorophore conjugate?**

Enzyme for maximum sensitivity on a single target, fluorophore for stable, quantitative and genuinely multiplexed detection. The choice follows the readout and the imager rather than the primary.

## Matrix and adhesion antibodies

Matrix proteins are cross linked, insoluble and often only detectable after conditions that destroy the epitope, while adhesion and junction proteins sit in complexes that a standard lysis buffer leaves in the pellet. Both classes fail for extraction reasons long before antibody quality becomes the issue.

1. **Decide native or denatured before buying**: Fibrillar matrix proteins present very different epitopes assembled and unfolded. A reagent validated on a denatured chain may see nothing in tissue, and one raised against the assembled form may see nothing on a blot.
2. **Extract for the compartment**: Cross linked matrix and junctional complexes are not solubilised by a standard buffer. Sequential extraction, or a strong denaturing buffer with the loss of native epitopes that implies, is what the blot requires.
3. **Choose retrieval per reagent, not per tissue**: Junction proteins in fixed tissue frequently need retrieval to be visible at all, and the optimum differs between reagents. Optimize on a tissue known to be positive rather than on the experimental material.
4. **Read junction stains as continuity, not intensity**: What matters at a junction is usually whether the staining is continuous along the contact, not how bright it is. Scoring intensity where continuity is the biology misses the effect.
5. **Confirm the matrix is yours**: Cultured cells sit on a coating and in serum containing medium, both of which contribute matrix protein. A no cell control on the same coating shows how much of the signal came with the plate.

**Why does my matrix protein not appear on a blot?**

Because it was not solubilised. Cross linked matrix stays in the pellet through standard lysis, and the sample that matters is the one most people discard. Sequential or denaturing extraction is required.

**Do junction proteins, such as a PECAM-1 antibody target, need retrieval?**

Usually, in cross linked fixed tissue, and the optimal conditions differ between reagents. A weak junction stain is far more often a retrieval problem than a reagent problem.

**How do I know the matrix signal is from my cells?**

Run the coating and the medium without cells through the same protocol. Serum and coatings contribute matrix proteins, and that background is frequently substantial in culture experiments.

## Lysosome and autophagy antibodies

Autophagy is a flux, and every marker in it is a snapshot of a compartment that is simultaneously being made and destroyed. A rise in a marker can mean more autophagy or blocked degradation, and the two are opposite conclusions. Only a comparison with and without a degradation inhibitor separates them.

1. **Always run the inhibitor arm**: The interpretable measurement is the difference between marker levels with and without a lysosomal inhibitor. A single measurement without that comparison cannot distinguish increased synthesis of autophagosomes from blocked clearance.
2. **Read the lipidated form as a ratio**: The informative quantity is the proportion of the marker in its membrane bound form, and it needs a gel that resolves the two species cleanly. Reporting only the lipidated band is a common and misleading shortcut.
3. **Use the cargo receptor as the second axis**: A receptor that is itself degraded by the pathway falls when flux increases and accumulates when it is blocked. Read in the opposite direction to the membrane marker, it constrains the interpretation considerably.
4. **Treat machinery levels as capacity, not activity**: Core machinery proteins are present whether or not the pathway is running. Their levels report capacity, and reading them as activity is one of the most frequent errors in this literature.
5. **Distinguish zymogen from mature protease**: Lysosomal proteases are made as inactive precursors and matured in the compartment. A reagent detecting both reports synthesis; only the mature band reports delivery to a functioning lysosome.

**Does more autophagosome marker mean more autophagy?**

Not on its own. It can equally mean autophagosomes are accumulating because they are not being cleared. The inhibitor comparison is what distinguishes induction from blockade, and no single measurement substitutes for it.

**Does a LAMP 1 antibody measure lysosome function?**

No. It counts compartments carrying that membrane protein. Function requires acidification and active proteases, which are measured with pH sensitive probes and by looking at protease maturation.

**Why does the cargo receptor, an SQSTM1 antibody target, go the other way?**

Because it is degraded along with the cargo it binds. Falling receptor with rising membrane marker indicates flux; both rising together indicates a block.

## Mitochondrial antibodies

Mitochondrial antibodies are used for two incompatible purposes: normalising to mitochondrial mass and reporting mitochondrial function. A marker that is a good mass reference is by definition insensitive to function, and a subunit whose level responds to metabolic state is a poor normaliser. Choosing without deciding which you need is the recurring error.

1. **Separate mass from function first**: If the question is how much mitochondria a cell has, use a marker known to scale with mass. If the question is respiratory capacity, use subunits and enzyme activities and do not normalise them to each other.
2. **Remember the two genomes**: Respiratory complexes contain subunits encoded in the nucleus and in the mitochondrion, and they respond differently to stress and to drugs. A cocktail covering both is far more informative than any single subunit.
3. **Match the reagent to the membrane**: Outer membrane, inner membrane, intermembrane space and matrix proteins behave differently in fractionation and in permeabilisation experiments. Naming the sublocation is part of choosing the reagent.
4. **Read shape proteins by imaging**: Fusion and fission machinery levels say little about network morphology. The measurement that reports shape is imaging with a segmentation analysis, and the blot is at best supporting evidence.
5. **Normalise carefully**: Normalising a mitochondrial protein to a cytosolic loading control confounds mitochondrial content with expression. Where content is the variable, report both the mitochondrial marker and the total protein and let the reader see the ratio.

**Which marker is best for mitochondrial mass?**

An abundant, stable structural protein rather than a respiratory subunit, because subunit levels respond to metabolic state. Confirm in your own system that the chosen marker tracks an independent measure of mass before relying on it.

**Why measure subunits from both genomes?**

Because nuclear and mitochondrially encoded subunits are regulated separately, and a mismatch between them is itself an informative finding about mitochondrial translation or genome maintenance.

**Can an OPA1 antibody blot report mitochondrial morphology?**

No. Fusion and fission protein levels are only loosely related to network shape, which has to be imaged and segmented. Reporting morphology from a blot is a claim the data does not support.

## Vesicle and tetraspanin antibodies

There is no marker that identifies a vesicle as having come from a particular compartment. The tetraspanins used routinely are enriched on small vesicles and are also present on the cell surface and on other particles, so a positive stain establishes enrichment rather than origin. Preparations are characterised by panels, including negative markers.

1. **Use at least three positive markers**: Single marker characterisation is not accepted, because each marker is present on a subset and on other material. A panel spanning different marker classes describes the preparation far better than any one of them.
2. **Include a negative marker**: An abundant intracellular protein from a compartment that should not be in the preparation shows whether cell debris or lysed material contaminates it. This is the control that most preparations lack and most reviewers ask for.
3. **Report the isolation method with every result**: Differential centrifugation, density gradient, size exclusion and precipitation yield different populations with different contaminants. Two preparations by different methods are not comparable, and the method belongs in the figure legend.
4. **Measure particle number and protein separately**: Protein per particle differs enormously between preparations and between methods. Reporting only protein makes a contaminated preparation look abundant.
5. **Do not infer origin from markers alone**: Marker enrichment supports a claim about vesicle type; it does not establish biogenesis route. Where origin matters, genetic perturbation of the pathway is what supports it.

**Is an exosome antibody stain, such as a CD63 antibody, proof of exosomes?**

No. Those markers are enriched on small vesicles and are also found on the plasma membrane and other particles. They support a claim of enrichment, and origin needs perturbation of the biogenesis pathway.

**Which negative marker should be used?**

An abundant protein from a compartment that should be absent, run on the same blot with equal loading. Its absence is what distinguishes a vesicle preparation from a preparation of cell fragments.

**Why do preparations differ so much between laboratories?**

Because isolation method, starting material and culture conditions all change what is recovered. Reporting the method in full is the minimum requirement for anyone else to interpret the result.

## Oncology target antibodies

An antibody against a tumour surface target is bought to answer one of two very different questions: is the target present, and how much of it is there. Presence tolerates most reagents; scoring does not, because it depends on the epitope surviving fixation and on the stain behaving linearly over the range being scored.

1. **Establish which name the target goes by**: Several of these targets have two or three names in common use, and catalogs index them differently. Confirm the gene identity before comparing reagents, because apparently different products are frequently the same target.
2. **Match the epitope to the application**: An extracellular epitope is what a live cell stain or a flow panel needs. An intracellular epitope may be the only one that survives fixation for tissue work. A reagent validated on lysate says nothing about either.
3. **Treat scoring as a quantitative assay**: If a stain will be used to grade expression, it needs controls spanning the range, a defined scoring rule and evidence that intensity tracks amount. Reagents sold for research are not calibrated for this.
4. **Watch for shed and soluble forms**: Several surface targets are shed, so a soluble measurement and a cell surface measurement can diverge. State which you measured and do not treat them as the same quantity.
5. **Control expression, not just staining**: A cell line panel spanning negative, low and high expression is the control that makes a stain interpretable, and it transfers across experiments better than any isotype control.

**Can a research antibody be used to score expression?**

For internal research comparisons with proper controls, yes. For anything informing a clinical decision it cannot, because scoring reagents are validated and labeled for that purpose and research reagents explicitly are not.

**Why does a reagent work on a blot and not on tissue?**

Different epitope accessibility. Denatured protein on a membrane presents regions that a cross linked tissue section hides, and retrieval has to be optimized per reagent rather than per tissue.

**Extracellular or intracellular epitope for a c-Met antibody?**

Extracellular for live cells, flow cytometry and anything measuring surface density. Intracellular where fixation has destroyed the surface epitope, accepting that the measurement is then total rather than surface protein.

## Neurodegeneration antibodies

In neurodegeneration the interesting species is usually the one that has left solution. Aggregated, phosphorylated and truncated forms are what the disease is about, and each needs a different reagent and a different extraction. A total protein antibody on a soluble lysate frequently reports the fraction that is not the point.

1. **Fractionate by solubility deliberately**: Sequential extraction into soluble and insoluble fractions is the core technique here, because the pathological species partitions into the fraction most protocols discard. Blot both and show both.
2. **Name the species the reagent detects**: Total, phosphorylated at a defined site, conformational and truncated reagents exist for the same protein and report different things. Buy against the species in the hypothesis, and say which in the methods.
3. **Preserve modification state at collection**: Phosphorylation is lost quickly post mortem and in a warm lysate. Handling time, temperature and inhibitor presence dominate the measurement of any phosphorylated species.
4. **Read glial markers as state dependent**: Markers used to identify glial populations change with activation, so a reduced signal can mean fewer cells or a changed state. Pair an identity marker with a state marker before drawing a conclusion about numbers.
5. **Confirm aggregates by more than a blot**: Insolubility on a blot is suggestive; a conformation specific reagent, a filter trap assay or imaging of inclusions is what supports an aggregation claim.

**Why is my target absent from the lysate?**

Because it went into the pellet. In this field the insoluble fraction is frequently the sample that matters, and a protocol that discards it will report a loss where there was a redistribution.

**Total or phospho reagent?**

Both, from the same sample. The phosphorylated species is only interpretable as a fraction of the total, and treatments and disease states change both.

**Does a glial marker such as a TREM2 antibody count cells?**

Only if the marker is state independent, which most are not. Activation changes expression, so an apparent change in cell number may be a change in phenotype. Use two markers with different regulation.

## Nucleic acid modification antibodies

An antibody raised against a chemical modification binds that mark wherever it occurs, which makes it powerful and unusually easy to misuse. Access to the mark frequently requires a denaturation or digestion step, the signal is a distribution rather than a band, and specificity has to be demonstrated against the unmodified template rather than assumed.

1. **Make the mark accessible**: Marks on double stranded nucleic acid are not available to an antibody until the template is denatured, and marks on folded proteins may need digestion or unfolding. The accessibility step is the method, and skipping it produces a clean negative.
2. **Prove specificity against the unmodified template**: Run a sample known to lack the modification and, where possible, a synthetic template carrying and lacking it. Without that pair the signal cannot be attributed to the mark.
3. **Enrich before detecting where the mark is rare**: Most modifications occupy a small fraction of their substrate. Immunoprecipitation with the modification reagent followed by detection of the substrate is frequently the only way to see it.
4. **Expect a smear and quantify it as one**: A mark present across many proteins or many sites gives a distribution. Integrating a region, with a defined region and a matched control lane, is the honest quantification.
5. **Block the erasers at collection**: Enzymes that remove these marks stay active in a lysate. Their inhibitors have to be in the buffer before the sample meets it, exactly as for phosphorylation.

**Why does my modification stain show nothing?**

Usually accessibility. Marks on double stranded nucleic acid need denaturation, and buried protein marks need digestion or unfolding. Add the step before changing reagents.

**How is specificity of a 5mC antibody demonstrated?**

With a template known to carry the mark and one known to lack it, run side by side, plus a competition with free modified substrate where available. A supplier's dot blot is a starting point, not evidence in your system.

**Can these reagents be quantitative?**

Relatively, within one experiment with matched controls and a defined integration region. Absolute quantification of a modification by antibody is not reliable, and mass spectrometry is the method when a number is needed.

## Kinase signalling antibodies

A kinase blot is only interpretable as a pair. The phosphorylated form reports activity, the total reports how much enzyme is there, and treatments routinely change both. Buying only the phospho reagent produces a figure nobody can normalise, and handling that loses phosphorylation produces a negative from a positive experiment.

1. **Buy phospho and total together, always**: The two reagents must be raised for the same protein and used on the same sample, ideally the same membrane through fluorescent multiplexing. A phospho signal without its total is uninterpretable.
2. **Name the site**: Kinases carry several regulatory phosphorylation sites that report different things, including activation, inhibition and priming. A reagent described only as phospho is not a specification.
3. **Put phosphatase inhibitors in first**: Phosphorylation is stripped within minutes in a cold lysate. Inhibitors go into the buffer before the cells meet it, and the path from plate to denatured sample must be short and rehearsed.
4. **Time course before single timepoint**: Kinase activation peaks in minutes. One time course per cell type and stimulus establishes where the peak is, and every subsequent single timepoint is chosen from that curve.
5. **Use an inhibitor as the specificity control**: A selective inhibitor of the kinase should collapse the phospho signal without changing the total. That pair of lanes is stronger evidence than any datasheet.
6. **Treat small GTPases as activity assays**: Nucleotide bound state, not abundance, is what matters for these proteins, and it is measured by pull down with an effector domain rather than by a blot of total protein.

**Why did my phospho signal disappear?**

Handling, in most cases: missing phosphatase inhibitor, warm buffer, or too long between the dish and denaturation. Fix the lysis protocol before suspecting the reagent.

**Which phosphorylation site should I measure?**

The one whose regulation your hypothesis concerns, named explicitly. Different sites on the same kinase report activation, inhibition and priming, and reporting phospho without a site is not a result.

**Can total kinase level from an AKT antibody substitute for activity?**

No. Expression and activity are independent, and treatments commonly change one without the other. Where activity is the claim, measure phosphorylation at a defined site or run an activity assay.

## Immune receptor CD antibodies

Surface marker antibodies are the most standardized reagents in biology and the easiest to use badly. Two clones against the same cluster can stain different proportions of the same sample, Fc receptors capture immunoglobulin regardless of specificity, and a dilution taken from a datasheet is almost never the right one for your cells.

1. **Block Fc receptors before staining**: Cells carrying Fc receptors bind antibodies through the constant region, producing signal in every channel. A blocking reagent appropriate to the species goes on before the panel, and its absence is the commonest cause of an uninterpretable myeloid stain.
2. **Record the clone, not the cluster**: A cluster designation names an antigen, not a reagent. Two clones can recognize different epitopes with different accessibility, and switching clone is a protocol change requiring re titration.
3. **Titrate on your own cells**: The optimal amount is the one giving the best separation between positive and negative populations on your sample, and it is usually below the datasheet dilution. One titration per clone pays for itself immediately.
4. **Match fluorophore brightness to antigen density**: Low density antigens need bright fluorophores on clean detectors; high density lineage markers can sit anywhere. Getting this backwards is why panels underperform their design.
5. **Check the epitope survives your processing**: Enzymatic dissociation cleaves several surface antigens, and fixation destroys others. Where a marker matters, confirm with an alternative clone or a gentler protocol.

**Why is everything positive in my myeloid stain despite a good CD32 antibody?**

Almost certainly unblocked Fc receptors. Add an appropriate blocking step before the panel and the background usually disappears entirely.

**Are two clones against the same marker interchangeable?**

No. They recognize different epitopes with different sensitivity to fixation and dissociation, and they frequently stain different fractions. Treat a clone change as a protocol change.

**Should I use the datasheet dilution?**

As a starting point only. Titration on your own cells almost always improves separation, and it is one plate of work per clone.

## MHC and presentation antibodies

Presentation molecules are the most polymorphic proteins there are, and a monoclonal antibody recognizes an epitope that some allele groups carry and others do not. A clone that stains one donor strongly and another weakly may be reporting genotype rather than expression, which is a serious problem when the claim is loss of presentation in disease.

1. **Establish what allele groups the clone covers**: Datasheets vary in how clearly they state this, and it is the single most important property. A pan reagent against a conserved framework region is safer than an allele specific one when total surface expression is the question.
2. **Use a partner chain as a cross check**: The invariant light chain of class I is not polymorphic, so a reagent against it gives an allele independent measure of surface class I. Disagreement between it and a heavy chain reagent is informative.
3. **Interpret loss with two independent reagents**: Apparent downregulation with one clone can be an allele the clone does not see. Two reagents recognising different regions, agreeing, is the minimum for a loss claim.
4. **Include an interferon treated positive control**: Presentation molecules are strongly induced by interferon, which gives a reliable high expression control and shows the stain has dynamic range on your cells.
5. **Separate surface from total**: Total protein can be unchanged while surface presentation collapses, because the block is in loading or trafficking. Surface staining of live cells answers the question that a permeabilised stain does not.

**Why do two donors stain so differently with an HLA-A antibody?**

Frequently genotype rather than expression. Clones recognize epitopes present in some allele groups and absent in others, which is why a framework or partner chain reagent is safer for comparing donors.

**How should loss of presentation be demonstrated?**

With two independent reagents, a surface rather than total measurement, and an interferon treated positive control on the same cells. One clone showing reduced staining is not sufficient.

**What is a good positive control?**

Interferon treatment, which strongly induces presentation molecules on most cell types. It calibrates the upper end of the stain and confirms the reagent detects induction.

## Tumour suppressor antibodies

In this area the intuitive reading is frequently backwards. A strong stain for a tumour suppressor often indicates a mutation that stabilises a non functional protein, and absence of a stain is weak evidence of loss because a weak stain has many causes. Mutation specific reagents exist for a handful of changes and are exquisitely dependent on the exact substitution.

1. **Know which direction the stain runs**: For proteins stabilised by mutation, high signal indicates a likely mutation and low signal indicates either normal turnover or a truncating change. The stain is a pattern, and the pattern has to be interpreted rather than read as amount.
2. **Treat absence as the hardest claim**: A negative stain can mean loss, a failed reagent, poor fixation or an epitope that was removed by a truncation. An internal positive control within the same section, such as normal stroma, is what makes absence credible.
3. **Use mutation specific reagents only for their exact change**: A reagent raised against one substitution detects that substitution and usually not its neighbors. Reporting it as detecting mutant protein generally is an overstatement.
4. **Confirm surrogates against sequence**: Where a stain is used as a proxy for a genotype, its concordance with sequencing has to be established in your own material before it is relied on.
5. **Separate lineage from driver**: Several reagents in this group identify a cell of origin rather than a driver event. Using them to infer mechanism is a common and avoidable overreach.

**Does strong staining mean the suppressor is working?**

Often the opposite. Missense changes that abolish function also abolish the degradation that keeps the protein low, so accumulation is a marker of a likely mutation rather than of activity.

**Can a stain replace sequencing?**

For a few well characterised changes with validated reagents, it can be an effective screen. It is not equivalent, and the concordance has to be established locally before the stain is used in place of the sequence.

**How do I make a negative anti p53 antibody stain credible?**

With an internal positive control in the same section, a reagent whose epitope is unlikely to be lost by truncation, and where possible a second reagent against a different region.

## Metabolic pathway antibodies

Several central metabolic regulators are controlled by destruction or by cleavage rather than by transcription, and the interesting form exists only under the condition being studied. Collect the sample wrongly and the target is gone before the lysate is made, which is why handling dominates this class of experiment.

1. **Lyse under the condition, not after it**: Oxygen sensitive factors are degraded within seconds of returning cells to air. Lysing inside the chamber, or with a denaturing buffer added immediately, is the difference between a band and nothing.
2. **Detect the cleaved form for processed factors**: Factors activated by proteolysis are inactive as the full length precursor. A reagent that sees only the precursor reports the reservoir, not the signal, and both bands should appear on the same blot.
3. **Extract membranes properly for transporters**: Multipass transporters are poorly solubilised by standard buffers and frequently aggregate when heated. Use a membrane protocol and load without boiling where the supplier recommends it.
4. **Distinguish enzyme amount from flux**: An enzyme level is not a rate. Where the claim is about metabolism, pair the blot with a functional measurement, because expression and flux frequently move independently.
5. **Control with a pathway agonist**: A treatment that induces the pathway calibrates the blot and shows the reagent detects the regulated form. For oxygen sensitive targets a chemical stabiliser is a convenient positive control.

**Why can I never see my hypoxia factor?**

Because it is destroyed almost immediately in oxygen. Lyse inside the hypoxic environment or add denaturing buffer instantly, and use a chemical stabiliser as a positive control to confirm the reagent works.

**Why does my transporter run as a smear with an SREBP1 antibody nearby?**

Glycosylation, aggregation on heating, or both. Load without boiling if the supplier advises it, and treat a broad band as the expected appearance rather than a failure.

**Does enzyme level report metabolic flux?**

No. Flux depends on substrate, allosteric regulation and modification as much as on amount. Pair the blot with a functional readout when the claim is about metabolism.

## Glial and neural identity antibodies

Markers in nervous tissue identify a cell only within a context, and several of the most used ones change with activation, injury or maturation. A reduced stain can mean fewer cells or the same cells in a different state, and distinguishing the two requires a second marker with different regulation.

1. **Pair a state independent marker with a state sensitive one**: Counting cells needs a marker that does not change with activation. Reporting state needs one that does. Using a single marker for both jobs is how apparent cell loss is reported where there was a phenotype change.
2. **Optimize retrieval per reagent**: Nuclear transcription factors in fixed tissue frequently need retrieval to be visible, and the optimum differs between reagents. Optimize on tissue known to be positive rather than on the experimental sections.
3. **Check the family, not just the member**: Several of these targets belong to families whose members share sequence and are expressed in other cell types. Read the immunogen and check the datasheet for family cross reactivity before assigning a cell type.
4. **Use morphology as a control on the marker**: In tissue, the shape and position of a stained cell is independent evidence. A marker signal in cells with the wrong morphology is a warning, not a discovery.
5. **Count with a defined rule and a blinded scorer**: Cell counting in sections is dominated by sampling and by scoring rules. Define the region, the section interval and the inclusion criteria in advance, and score blind.

**Does a reduced marker mean cell loss?**

Not on its own. Activation, injury and maturation all change marker expression, so a second marker with different regulation, or a nuclear count, is what distinguishes loss from a state change.

**Why is my transcription factor stain weak?**

Retrieval, usually. Nuclear targets in cross linked tissue need conditions optimized per reagent, and a weak nuclear stain is far more often a retrieval problem than a reagent problem.

**Can one marker such as a DCX antibody define a glial population?**

Rarely. Family members are shared with other cell types and expression shifts with state. Two markers plus morphology is the practical minimum for a population claim.

## Synaptic antibodies

A synaptic marker produces puncta, and a punctum on its own is not a synapse. The measurement that means something is apposition between a presynaptic and a postsynaptic marker, resolved well enough to be believed. Meanwhile the receptors and channels in the same tissue are membrane proteins that defeat a standard lysis buffer.

1. **Score apposition, not puncta**: Count pairs where a presynaptic and a postsynaptic marker are adjacent within the resolution limit. Counting one marker alone measures protein clusters, some of which are not synapses.
2. **Fix for the compartment**: Synaptic scaffolds and membrane receptors have different fixation optima, and a protocol that preserves one can destroy the other. Where both are needed, test the compromise on control tissue first.
3. **Extract membrane targets deliberately**: Channels and receptors are multipass membrane proteins that stay in the pellet with standard lysis and aggregate when boiled. Use a membrane protocol and follow the supplier's loading advice.
4. **Separate transmitter identity from synapse number**: Vesicular transporters assign the transmitter; scaffolds count contacts. Reporting a change in one as a change in the other is a frequent overreach.
5. **Control the resolution claim**: Apposition at the diffraction limit is suggestive rather than conclusive. Where the claim depends on it, use a super resolution method or an orthogonal measurement such as electrophysiology.

**Is a punctum a synapse?**

Not necessarily. Scaffold and vesicle proteins form clusters that are not all synaptic. Apposition of pre and postsynaptic markers is the standard evidence, and even that is limited by resolution.

**Why can I not detect my receptor on a blot?**

Because it stayed in the pellet or aggregated on heating. Multipass membrane proteins need a membrane extraction protocol and usually gentler denaturation than a standard sample.

**Which marker, such as a VAChT antibody, assigns transmitter identity?**

The vesicular transporter, because it determines what the terminal releases. Synthetic enzymes are informative and less definitive, since the enzyme can be present where the transporter is not.

## Muscle and plasma protein antibodies

Two problems dominate this group. Muscle proteins exist as tissue specific isoforms that share most of their sequence, so a reagent that does not distinguish them reports the wrong tissue. Plasma proteins are abundant enough to saturate everything, and their measurement is dominated by collection and by dilution rather than by the antibody.

1. **Name the isoform for muscle targets**: Cardiac, skeletal and smooth muscle isoforms of the same protein family differ in a small number of residues. A reagent that does not state which it detects cannot support a tissue specific claim.
2. **Extract muscle properly**: Very large and highly cross linked muscle proteins need specialized gels and extraction, and standard sample preparation leaves them at the top of the well. Use a protocol designed for the size range.
3. **Treat abundant plasma proteins as a dilution problem**: Highly abundant analytes saturate immunoassays and mask everything else. Dilution series, depletion where appropriate and a clear statement of the dilution used are what make these measurements comparable.
4. **Standardize blood collection**: Tube type, time to processing, temperature and freeze thaw change measured concentrations, sometimes by more than the biological effect. Fix the collection protocol before the first sample.
5. **Do not assume a loading control is stable**: Common structural loading references change with differentiation, stress and tissue type. Total protein staining is safer, and should be shown.

**Does a troponin reagent distinguish cardiac from skeletal?**

Only if it says so. The isoforms differ at defined regions and clones vary in whether they discriminate. For any tissue specific claim the datasheet must state it explicitly.

**Why does my very large muscle protein not enter the gel?**

Standard acrylamide gels do not resolve proteins of that size. Low percentage or gradient gels designed for the range, and a gentler extraction, are required.

**How much does blood collection matter for a BNP antibody assay?**

Frequently more than the biology. Tube additive, time to processing and freeze thaw all change measured concentrations for several plasma analytes, which is why a fixed protocol comes before the first sample.

## Bioanalytical antibody reagents

Measuring a therapeutic antibody in a sample full of immunoglobulin is a specificity problem that only an idiotype specific reagent solves. Measuring the response against that therapeutic is harder still, because the drug itself interferes with the assay designed to detect antibodies against it. Both problems are reagent problems before they are assay problems.

1. **Commission the idiotype pair early**: Two reagents recognising different regions of the variable domain, one for capture and one for detection, are what make a quantitative assay possible. They take months to generate and sit on the critical path of every study that needs them.
2. **Design for drug tolerance**: Circulating drug binds the antibodies you are trying to detect. Acid dissociation and other pretreatments are what let the assay see them, and the tolerance achieved has to be measured and reported.
3. **Confirm neutralisation functionally**: A binding assay shows that an antibody response exists; a cell based assay shows whether it blocks activity. They answer different questions and regulators expect the second where the first is positive.
4. **Fix the concentration method**: Absorbance, colorimetric assay and immunoassay give different numbers for the same material. Pick one, state it, and use it consistently, because concentration errors propagate into every downstream calculation.
5. **Validate against a written protocol**: Accuracy, precision, selectivity, dilutional linearity, stability and a stated range, agreed before the assay is used. An assay validated after a surprising result is validated to explain it.

**Why does an anti-idiotype reagent take so long to make?**

Because it has to bind one variable region and nothing else, which requires immunisation or selection against that molecule and extensive counter screening. It is a discovery project, not a catalog purchase, and it belongs early in a program's plan.

**What is drug tolerance and why does it matter?**

It is the concentration of circulating drug at which the assay can still detect antibodies against it. Without adequate tolerance the assay reports negatives from samples that are positive, which is the most consequential failure in this area.

**Binding or neutralising cell based assays, and in what order?**

Both, in sequence. Binding assays screen; neutralising assays characterise the positives functionally. Reporting a binding result as evidence of neutralisation is not supported.

## Therapeutic antibody development

How human an antibody is affects how a patient's immune system treats it, and there are several routes to the same endpoint with very different timelines and freedom to operate. The decision is usually made early, on the basis of what platform is available, and its consequences appear much later in immunogenicity and in manufacture.

1. **Decide the route from where the lead came**: A rodent lead can be grafted onto human frameworks and back-mutated until affinity is recovered. A lead from a human library or a transgenic animal is already human sequence. Neither route guarantees low immunogenicity, which depends on sequence liabilities as well as origin.
2. **Budget the back-mutation cycle**: Grafting frequently loses affinity, and recovering it means restoring selected framework residues and re-testing. That cycle is the real cost of humanisation and it is routinely under-planned.
3. **Screen for developability, not only affinity**: Aggregation, viscosity at high concentration, chemical liabilities in the variable region and expression level decide whether a candidate can be made and formulated. Screening for them early is far cheaper than discovering them at scale up.
4. **Decide the format before the sequence is locked**: Bispecific and conjugated formats impose chain pairing, engineering and conjugation site requirements that are much easier to design in than to retrofit.
5. **Check freedom to operate on the platform**: Discovery platforms, humanisation methods, frameworks and formats all carry commercial terms that follow the molecule. Establish them before a lead is nominated.
6. **Characterise to a written panel**: Identity, purity, charge and size variants, glycosylation, potency and stability. Agreeing the panel early means the data accumulates as the program runs rather than being generated in a rush.

**Does a fully human sequence guarantee low immunogenicity?**

No. Sequence origin reduces one risk; aggregation, sequence liabilities, the target itself and patient factors all contribute. Immunogenicity is assessed, not assumed from provenance.

**Monoclonal or polyclonal for monoclonal antibody generation?**

Monoclonal for anything requiring reproducibility across lots and across years, which includes every therapeutic and most assays. Polyclonal remains useful where signal amplification across multiple epitopes matters and lot variation is tolerable.

**When should the format be fixed?**

Before the sequence is locked. Bispecific pairing and conjugation site engineering change the molecule, and retrofitting them means re-optimizing a candidate that had already been chosen.

**What does developability screening cover?**

Expression level, aggregation propensity, thermal stability, viscosity at formulation concentration and chemical liabilities such as oxidation and deamidation sites in the variable region.

## Pathogen and serology antibodies

Reagents in this area answer two different questions. Some detect a pathogen component directly; others detect the host response to it, which is evidence of exposure at some point rather than of current infection. Confusing the two produces conclusions the assay cannot support, and variant coverage adds a further layer for rapidly changing organisms.

1. **Separate antigen detection from serology**: Detecting a pathogen protein indicates the organism is present. Detecting host antibodies indicates exposure, possibly long past, and possibly from vaccination. State which the assay does.
2. **Ask about variant coverage**: For organisms that change quickly, a reagent raised against one variant may bind others weakly or not at all. Coverage has to be stated and, for anything important, tested against the variants in circulation.
3. **Handle samples at the right containment**: Clinical and environmental samples carry infectious risk that depends on the agent. Containment level, waste route and vaccination requirements are decided before the first sample arrives.
4. **Use immunogen and carrier reagents as controls**: Reagents against common carriers and model antigens are experimental tools: they calibrate an assay, check a conjugation and provide a negative. They are not diagnostics and should not be presented as such.
5. **Consider anti carrier responses in formulated products**: Responses against polymer carriers used in formulation are measurable and clinically relevant, and reagents against them belong in the bioanalytical package for those products.

**Does a positive serology result mean current infection?**

No. It indicates exposure at some point, which may include vaccination. Current infection is established by detecting the organism or its components, or by a rise in titre between paired samples.

**Will an RBD antibody raised against one variant detect others?**

Not reliably. Coverage depends on where the epitope sits and how conserved it is, and it has to be stated by the supplier and ideally tested locally against circulating variants.

**Why do assays include antibodies against carrier proteins?**

As controls and tools: to check a conjugation worked, to provide a negative, or to measure a response against a carrier used in a formulation. They are not diagnostic reagents.

## Autoantibody and complement reagents

A catalog listing can mean two entirely different things here: a reagent you buy to detect a protein, or a patient antibody that is itself the thing being measured. The same structural target appears in both roles, and the distinction decides the assay, the controls and what may be claimed from a result.

1. **State whether the antibody is reagent or analyte**: A reagent detects a target; an autoantibody is the target. Assays, controls and regulatory expectations differ completely, and the same protein name appears in both contexts.
2. **Read pattern before titre in autoantibody work**: Classical autoantibody testing on a cell substrate reports a staining pattern that suggests a family of specificities, followed by a confirmatory specific assay. The pattern is the screen, not the answer.
3. **Detect complement deposition in tissue, not plasma**: Deposited fragments in tissue are the informative finding in transplant and immune complex disease. Measuring circulating complement answers a different question about consumption and synthesis.
4. **Choose structural reagents for the compartment**: Nuclear body, nucleolar, centromere and nuclear pore reagents each give a characteristic pattern that is itself a control. A stain that does not look like the expected structure is a warning.
5. **Respect the labeling boundary**: Reagents sold for research use may not be used to produce a result that informs clinical care, however capable they are. Where the result will inform care, the assay has to be appropriate to that use.

**Is a listing for laboratory reagents against an autoantibody something I can buy?**

Sometimes both exist under the same name: a laboratory reagent against the structure, and an assay that measures a patient's antibodies against it. Read the product description carefully, because the two are not interchangeable.

**What does a staining pattern tell you?**

It narrows the likely specificity to a family and directs which confirmatory assay to run. It is a screening result, and reporting it as a specific diagnosis is not supported.

**Why measure deposited complement rather than circulating?**

Because deposition in tissue localises the process, which is what matters in transplant rejection and immune complex disease. Circulating levels report consumption and synthesis, which is a different question.

## Protease and remodelling antibodies

Proteases in this family are secreted as inactive precursors and activated by cleavage, and they are held in check by inhibitors that circulate with them. A blot of total protein therefore reports how much enzyme exists, which is a poor proxy for how much is working, and the difference is usually the point of the experiment.

1. **Resolve zymogen from active form**: The two differ in size and both are visible on a well resolved gel. Quantifying the total band as activity is the most common error in this area and it is avoidable by looking at the right band.
2. **Use a zymography or activity assay for activity**: Gelatin zymography and fluorogenic substrate assays measure what the enzyme does. They should accompany any claim about activity, because a blot cannot make it.
3. **Measure the inhibitor alongside**: Net activity depends on the enzyme to inhibitor ratio, and treatments frequently change both. Reporting the protease alone leaves the interpretation open.
4. **Account for complexes**: Several of these proteins circulate bound to partners, which changes their apparent size and can mask epitopes. A band at an unexpected size may be a complex rather than an artefact.
5. **Measure sheddase substrates in the medium**: When a sheddase acts, its substrate leaves the cell surface. Measuring the released fragment in the supernatant is a more direct readout of sheddase activity than blotting the enzyme.

**Does more protease mean more degradation?**

Not reliably. Most is present as inactive precursor and inhibitors are usually in excess. Net activity requires resolving the active form and measuring the inhibitors, or running an activity assay.

**What does zymography add over an MMP2 antibody blot?**

It reports enzymatic activity resolved by size, which a blot cannot. Its limitation is that the gel conditions can activate latent enzyme, so it is read as a comparison between lanes rather than as an absolute.

**Why is my band at the wrong size?**

Frequently a complex with an inhibitor or a carrier, which is real biology rather than an artefact. Run a reducing and a non reducing sample to see whether the shift is disulphide dependent.

## Lineage factor antibodies

Nuclear factors are low abundance, tightly bound to chromatin and frequently invisible in a whole cell lysate made the ordinary way. Most disappointing stains in this group are extraction and retrieval failures rather than reagent failures, and several of the markers report a transient state rather than a stable identity.

1. **Extract the nucleus, do not lyse the cell**: A standard whole cell buffer leaves much of a chromatin bound factor in the pellet. A nuclear extraction protocol, with the pellet checked, is what puts these targets on a blot.
2. **Optimize retrieval per reagent in tissue**: Nuclear epitopes in cross linked tissue almost always need retrieval, and the optimal buffer and duration differ between reagents. Optimize on tissue known to be positive.
3. **Separate identity from immediate early activity**: Several factors here are induced within minutes of any stimulus, including handling. Treating them as lineage markers produces conclusions about identity from what is actually a stress response.
4. **Score nuclear localisation, not intensity alone**: For factors that shuttle, the informative measurement is the proportion in the nucleus per cell. Whole field intensity mixes that with expression level.
5. **Include a known positive population**: A tissue or cell line known to express the factor, on the same slide or blot, is what makes a negative result credible and a weak result interpretable.

**Why is my nuclear factor, a NANOG antibody or SOX2 antibody target, absent from the lysate?**

Because it is bound to chromatin and stayed in the pellet. Use a nuclear extraction protocol and blot the pellet as well as the supernatant to confirm where the target went.

**Are immediate early genes useful as markers?**

As activity readouts, yes; as identity markers, no. They respond within minutes to handling, dissociation and medium change, so a difference between groups can be a difference in processing.

**How much does retrieval matter?**

More than the antibody, for most nuclear targets in fixed tissue. A weak nuclear stain is far more often under-retrieved than under-sensitive.

## Chromatin enzyme antibodies

Chromatin enzymes work inside complexes, and their catalytic subunit is present whether or not the complex is assembled and active. Detecting the enzyme therefore reports potential rather than activity, and the readout that reports activity is the mark on the substrate, measured with a different reagent and a different protocol.

1. **Measure the mark to measure the activity**: The enzyme level reports capacity. The abundance of the mark it deposits, and its distribution across the genome, report what the enzyme actually did. Both belong in a chromatin figure.
2. **Treat the complex as the unit**: Catalytic subunits are inactive alone and their targeting depends on partners. Where the claim concerns targeting, immunoprecipitation of the complex is the informative experiment rather than a blot of the subunit.
3. **Use histone variants as their own controls**: Variant specific reagents have to distinguish sequences that differ by a few residues. Demand the supplier's discrimination data and, where a conclusion depends on it, confirm with an orthogonal method.
4. **Normalise chromatin experiments to total histone**: Modification levels are reported relative to the histone carrying them, not to a cytoplasmic loading control. Blot the total histone on the same membrane.
5. **Validate chromatin immunoprecipitation reagents separately**: An antibody that works on a blot may not work in chromatin immunoprecipitation, where the epitope is in cross linked, native chromatin. Application specific validation is required rather than assumed.

**Does more enzyme mean more modification?**

Not reliably. Activity depends on complex assembly, targeting and substrate availability. Measuring the mark itself is what reports the outcome, and both measurements together are what support a mechanistic claim.

**Can a BRD4 antibody blot reagent be used for chromatin immunoprecipitation?**

Only if it has been validated for it. The epitope in cross linked native chromatin is presented very differently from a denatured band, and many reagents work in one and not the other.

**How should modification blots be normalised?**

To the total amount of the histone carrying the mark, on the same membrane. Normalising to a cytoplasmic protein confounds the modification with the amount of chromatin loaded.

## Receptor kinase antibodies

Receptors are membrane proteins, usually glycosylated, frequently multipass, and sometimes present in a rearranged form that keeps only part of the protein. Each of those properties defeats a standard protocol in a different way, and the reagent has to be chosen against the specific obstacle rather than against the gene name.

1. **Choose the domain the fusion retains**: Where a rearrangement is the target, the reagent must bind a region present in the fusion protein. An antibody against the lost portion will be negative in exactly the samples of interest.
2. **Extract membranes, do not boil them**: Multipass membrane proteins aggregate when heated and stay in the pellet with a standard buffer. Use a membrane protocol and follow the supplier's advice on denaturation temperature.
3. **Expect glycosylation to broaden the band**: Heavily glycosylated receptors run as smears and at sizes well above the predicted mass. A deglycosylation control confirms the identity of an unexpected band.
4. **Treat multipass receptors as the hardest class**: Reagents against seven transmembrane receptors have a poor reputation for specificity. Where a conclusion depends on one, a knockout or knockdown control is not optional.
5. **Separate expression from activation**: Receptor level and receptor phosphorylation are different measurements needing different reagents. Report both when the claim is about signalling rather than about presence.

**Why is my RET antibody receptor blot at the wrong size?**

Glycosylation, usually, which raises apparent mass and broadens the band. A deglycosylation treatment collapsing the band to the predicted size confirms the identity.

**Why are GPCR antibodies distrusted?**

Because many published reagents have failed knockout validation. It is a well documented problem in the field and it means a genetic control is required for any claim resting on one.

**How do I detect a fusion protein?**

With a reagent against a region retained in the fusion, ideally combined with an orthogonal method such as sequencing or a break apart assay. The domain choice is the whole question.

## Pipetting equipment

Pipetting is where most laboratory variability is created, and most of it comes from using a device outside its comfortable range or on a liquid it was not designed for. The devices differ by mechanism rather than by brand, and matching mechanism to liquid removes a category of error that no amount of care compensates for.

1. **Choose by volume decade, not by maximum**: Accuracy is worst at the bottom of any device's range. Cover the laboratory's real volumes with two or three ranges rather than one general purpose instrument used everywhere, and keep the routine volume near the middle.
2. **Match the mechanism to the liquid**: Air displacement suits aqueous liquids and fails on viscous, volatile and dense ones. Positive displacement handles those correctly. Using the wrong mechanism produces a systematic error nobody sees.
3. **Choose plate devices by format, not channel count**: Head geometry has to match the plate, including well spacing and depth. A head that will not reach the bottom of a deep well, or that fouls a skirt, is unusable whatever its channel count.
4. **Add repeat dispensing where the protocol repeats**: A repeater or a dispenser removes the aspirate and dispense cycle for a series of identical volumes, which cuts both time and variability. It is the cheapest automation in a laboratory.
5. **Calibrate on a schedule and keep the certificates**: Set the interval from usage and consequence, run a simple gravimetric check between formal calibrations, and file the certificates against the instrument identifier rather than in a drawer.

**Graduated or micropipette?**

A graduated glass or plastic pipette with a controller handles millilitre transfers where a small error is immaterial. A micropipette handles microlitre volumes where it is not. Using either outside its decade is the commonest source of avoidable error.

**When is positive displacement required?**

For viscous, volatile, dense and foaming liquids, where the air cushion in a standard pipette expands, contracts or lets liquid escape. The tips cost more and the measurement is correct.

**Is a repeat dispenser worth it?**

For any protocol that dispenses the same volume more than a handful of times, yes. It reduces both time and variability, and it is far cheaper than a liquid handling platform.

**How often should devices be calibrated?**

By use and consequence rather than by calendar alone, with an in house gravimetric check between formal calibrations to catch drift early.

**When is a repeat dispenser worth buying?**

When a protocol distributes the same aliquot many times, such as filling plates or dispensing reagent into tubes. Compare cost per dispense rather than cost per tip, because the consumable is a syringe-style tip rather than a standard one.

## Plasmid preparation

Plasmid preparation looks like one technique at four scales and is really two products. Small preparations exist to screen colonies and be discarded. Larger preparations feed transfections and animals, where endotoxin, supercoiled fraction and consistency matter, and where the column chemistry that suits screening starts to fail.

1. **Size the culture to the copy number**: A low copy or large construct yields far less per millilitre, and the column will not compensate. Scale the culture rather than overloading a column, which reduces both yield and purity.
2. **Match the grade to the downstream cells**: Screening tolerates anything. Transfection of primary or immune cells needs endotoxin control, and animal work needs more. Ordering the cheap grade for a sensitive application produces a biological response nobody attributes to the DNA.
3. **Specify the supercoiled fraction**: Mass says nothing about the fraction that is intact supercoiled plasmid, which is what transfects. Two preparations at the same concentration can perform very differently.
4. **Sequence before scaling**: Whole plasmid sequencing costs little and catches rearrangements that a restriction check misses. Doing it before a large preparation avoids making milligrams of the wrong construct.
5. **Buy the large preparations where labor dominates**: At milligram scale the consumables and the hands on time frequently exceed a service quotation, and the service returns a certificate. Price the labor before defending the in house route.

**Why is my plasmid DNA isolation kit yield low?**

Copy number, construct size, culture density or an overloaded column, in that order. Scaling the culture and respecting the column's stated capacity fixes most cases.

**Does endotoxin matter for transfection?**

For robust immortalised lines, often not. For primary cells, immune cells and anything in animals it matters considerably, and the response it provokes is easily mistaken for an effect of the construct.

**When should preparation be outsourced?**

At milligram scale, when a defined endotoxin level is required, or when the labor displaces more valuable work. Below that, kits are simple and cheap.

**Is a plasmid DNA miniprep kit different from a plasmid isolation kit?**

Not usually. The names describe the same silica or anion exchange chemistry at the smallest scale, and the differences that matter are the column's binding capacity, whether the protocol includes an endotoxin reducing wash, and the elution volume. Read the capacity and the elution volume rather than the name on the box.

## HPLC consumables

A validated chromatographic method assumes the consumables it was developed with. Columns from different manufacturers with the same nominal chemistry differ in bonding density, end capping and silica, and a vial septum or an unfiltered mobile phase introduces peaks nobody ordered. This page is about the parts that are treated as interchangeable and are not.

1. **Treat an equivalence chart as a shortlist**: Cross reference charts identify columns with similar chemistry, and similar is not the same. Any substitution in a validated method requires demonstrating equivalent performance on your own separation.
2. **Record the column identity and its history**: Manufacturer, part number, dimensions, particle size and serial, plus an injection and back pressure log. It is what turns a gradual decline into a scheduled replacement rather than a failed batch.
3. **Filter mobile phases and samples**: Particulates raise back pressure and shorten column life, and unfiltered samples are the commonest cause of an early column failure. A syringe filter compatible with the solvent is a small cost against a column.
4. **Choose vials, inserts and septa deliberately**: Septum material can leach and can core, glass type affects adsorption of basic compounds, and an insert changes the minimum draw depth. All three are method parameters for trace work.
5. **Know the column's void volume**: Retention is interpreted relative to the unretained time, and it changes with column dimensions and particle architecture. Calculating it makes gradient methods transferable between column formats.

**Is one analytical column in HPLC interchangeable with another of the same chemistry?**

Not reliably. Bonding density, end capping and the underlying silica differ between manufacturers and even between production lots, and selectivity follows. Any swap in a validated method needs performance data.

**Do vials really affect results?**

For trace analysis and basic compounds, measurably. Adsorption to glass and leaching from septa both show up at low concentrations, which is why the vial specification belongs in the method.

**How long should a column last?**

It depends entirely on sample cleanliness and mobile phase. Logging injections and back pressure per column turns replacement into a planned event rather than a mid batch surprise.

**How much does an analytical column HPLC method depend on the lot?**

Enough to matter at the margins and not enough to change a chemistry. Retention shifts by a few percent between lots of the same column, which a method with sensible windows absorbs and a method that was tuned to one lot does not. The practical protection is a system suitability check at the start of every sequence, not buying the same lot for ever.

## Preparative chromatography

An analytical separation asks whether two things can be resolved. A preparative separation asks how much can be loaded before they stop being resolved, and how much of the target is recovered. Those are different questions, and a method that is excellent analytically frequently scales badly because it was optimized for resolution rather than for capacity.

1. **Measure loading capacity before scaling**: Increase the load on the analytical column until resolution degrades to the acceptable limit. That point, scaled by column volume, tells you what the preparative run can carry, and it is usually lower than expected.
2. **Optimize for recovery, not for peak shape**: A preparative method is judged on how much pure material comes out. Overlapping fractions, adsorption losses and on column degradation all reduce recovery and none show in an analytical chromatogram.
3. **Choose the mechanism for the impurity that matters**: Reversed phase, hydrophobic interaction, ion exchange, affinity and hydroxyapatite each separate on a different property. The train is built around the impurity you have to remove, not around a default.
4. **Plan fraction collection and analysis**: Fractions have to be triggered, collected, analyzed and pooled, and the analytical method for that has to exist before the preparative run. Collecting into tubes with no plan is how material is lost.
5. **Consider continuous operation for repeated separations**: Where the same separation runs many times, continuous or simulated moving bed operation raises productivity and reduces solvent substantially, at the cost of a more complex system and method.

**Why does my method fail on an HPLC chromatography machine at preparative scale?**

Overloading, usually. Analytical methods are developed at loads far below capacity, and resolution collapses as the load rises. Measuring the loading limit first is the step that prevents this.

**In HPLC liquid chromatography, is a bigger column always the answer?**

Only up to the point where packing quality, flow distribution and pressure become limiting. Beyond that, a different mechanism or a continuous approach usually beats a larger column.

**How is recovery measured?**

Mass of pure target out divided by mass in, with purity defined by the analytical method. Reporting yield without purity, or purity without yield, describes half the result.

## Contract testing services

Every result from a contract laboratory is the product of a method, a sample and a scope of accreditation. Buyers compare price and turnaround, which are the two least discriminating properties, and discover the method mismatch when a report is rejected by whoever asked for it.

1. **Name the method by its standard**: Different standards for the same property give different numbers. State the standard in the enquiry, and check it appears on the laboratory's published accreditation schedule rather than on its website.
2. **Get the sampling right first**: For microbiological and residue work the sample dominates the result. Sampling plan, container, preservation and time to the laboratory are decided before collection and are frequently the reason a result is unusable.
3. **Ask what recovery the method achieves**: A limit of detection assumes a recovery from your matrix. Ask for recovery data on a matrix like yours, because the published figure was generated on a different one.
4. **Treat failure analysis as an investigation**: Failure work delivers a reasoned opinion supported by evidence rather than a measurement. Agree the scope, the techniques and what happens if the evidence is inconclusive, before the work starts.
5. **Specify what the report must contain**: Method reference, sample description, date and condition on receipt, result with uncertainty, any deviations and a signature. Agree it in advance rather than discovering an omission when the report is needed.

**Why did two laboratories give different results?**

Different standards, different sample preparation or different recovery from your matrix. Naming the standard and asking for matrix recovery data resolves most discrepancies before they occur.

**How much does sampling matter?**

For microbiological and residue testing, frequently more than the analysis. A perfect method on an unrepresentative or badly preserved sample produces a precise wrong answer.

**What does a failure analysis deliver?**

An evidenced opinion about the mechanism, supported by the techniques applied. It is not a measurement, and the scope, the techniques and the limits should be agreed in writing at the start.

**Why does an enumeration method need validating?**

Because preservatives and the product's own chemistry suppress growth. The method must be shown to recover a known low inoculum in the presence of the product, or the count may be reporting inhibition rather than content.

**Does a negative sterility test prove the batch is sterile?**

No. It shows nothing grew in the sample tested; the sampling plan is what connects that to the batch. Its worst failure is a false positive from the environment, which is why the test is run in an isolator.

**Who needs microbiological testing of food, and how often?**

Anybody placing food on the market needs a plan proportionate to the hazard, which is a matter of the process rather than the size of the business: a ready to eat product with a long shelf life carries more testing than a product that is cooked by the customer. Frequency follows the same logic, and it is set by how quickly the process could drift without anybody noticing rather than by a calendar.

## Analytical instruments

Analytical instruments are bought by technique and should be bought by the question: which element or property, in which matrix, at what level, with what interferences present. Two techniques that measure the same element can report different values for good reasons, and the specification has to name the method for the number to travel.

1. **Start from element, matrix and level**: The required detection limit in your actual matrix, not in a clean standard, selects the technique. A method adequate for a percentage level determination will not reach trace levels and vice versa.
2. **Map the interferences before choosing**: Spectral and isobaric interferences are what separate techniques in practice. Ask each vendor how they handle the specific interferences your matrix presents, with data on a similar sample.
3. **Cost the sample preparation**: Digestion, combustion, fusion or extraction usually costs more time than the measurement and determines the achievable accuracy. Preparation equipment and its consumables belong in the comparison.
4. **Separate bulk from surface and structure**: Bulk composition, surface composition and crystal structure are different measurements needing different instruments. Buying one and inferring the others is a frequent and expensive assumption.
5. **Plan calibration and reference materials**: Every technique here needs certified reference materials matched to the matrix, run on a schedule. Their availability and cost is part of the running cost and occasionally the constraint.

**Which technique for trace elements?**

It depends on the element, the matrix and the required limit. Mass detection generally reaches lower levels and suffers different interferences from optical emission, so the choice is made against your matrix rather than in general.

**Why do two analytical lab instruments disagree?**

Different sample preparation, different calibration standards or an unaddressed interference. Harmonising preparation and running a matched certified reference material resolves most disagreements.

**Can one instrument cover bulk and surface?**

No. Surface sensitive techniques probe a very thin layer and bulk techniques average through the sample. They answer different questions and both are needed when a coating or a contamination layer is involved.

## Microscope configurations

Microscope configuration is decided by where the sample sits, what illuminates it and what has to be measured. Those three answers select the stand, the illumination path and the detector, and getting them wrong produces an instrument that works beautifully on the wrong kind of sample.

1. **Let the vessel choose the stand**: Cells in a flask or plate need an inverted stand with long working distance optics. Slides need an upright. Opaque samples need reflected illumination. This is the first decision and it is not adjustable later.
2. **Specify illumination as a path, not a lamp**: Excitation source, filter sets, dichroic and emission filters together determine what you can see. A stand with the wrong filter set for your fluorophores is unusable regardless of the objectives.
3. **Decide whether the instrument measures or shows**: Any microscope used for measurement needs calibration against a traceable graticule per objective and camera, on a schedule. Instruments used for illustration do not, and conflating them produces numbers nobody can defend.
4. **Buy screening instruments with their analysis**: An automated imaging platform generates data faster than a laboratory can analyze it. The analysis pipeline, the storage and the person are part of the purchase, and usually the larger part.
5. **Specify ergonomics for heavy use**: Eyepiece height, stage position and camera based viewing decide whether an instrument can be used for hours. Where it will be, this belongs in the specification alongside the optics.

**Inverted or upright?**

Inverted for cells in vessels and for anything requiring access from above; upright for slides and for reflected light work on opaque samples. It is decided by the sample and it cannot be changed afterwards.

**What does multiphoton microscopy add over confocal?**

Depth penetration in scattering tissue and less out of focus photodamage, at higher cost and complexity. For thin samples a confocal is simpler and usually better.

**Does a screening platform replace a microscope?**

No. It answers questions about many wells at moderate resolution; a research microscope answers questions about few samples at high resolution. Most laboratories that do both need both.

## Flow cytometry reagents

A cytometry panel is a negotiation between the fluorophores available, the lasers and filters installed, and the antigens being measured. Reagents chosen without reference to the instrument configuration produce a panel that cannot be resolved, and blocking that is skipped produces signal in every channel.

1. **Write the instrument configuration down first**: Lasers, filters and detectors define which fluorophores are excitable and separable. Every panel design starts from that table, and designing without it wastes reagents.
2. **Block Fc receptors before staining**: Cells carrying Fc receptors bind antibody through the constant region and produce signal in every channel. Blocking is a required step for myeloid and mixed populations, not an optional refinement.
3. **Assign bright fluorophores to dim antigens**: Fluorophore brightness on your configuration, matched against antigen density, is what determines resolution. Getting this backwards is the commonest reason a well designed panel underperforms.
4. **Include full minus one controls**: For every boundary that has to be drawn, a control lacking one color from the full panel shows where spread from the other channels puts the edge. Isotype controls do not address spread.
5. **Track instrument performance daily**: Bead based performance tracking, recorded over time, is what distinguishes an instrument drift from a biological change. It takes minutes and it is the reference when a result surprises you.
6. **Budget beads, service and the operator**: Setup and tracking beads, a service contract and a trained operator are the recurring costs. On a shared instrument the operator is the difference between reliable data and an expensive queue.

**Can I use any fluorophore combination?**

Only those your lasers can excite and your filters can separate. The instrument configuration is the constraint, and panel design tools are only useful once that configuration is entered correctly.

**Is Fc block flow cytometry always necessary?**

For any sample containing cells with Fc receptors, which includes most primary immune preparations. Skipping it produces plausible looking positive signal that is not specific.

**What controls does a high parameter panel need?**

Single stain compensation or unmixing controls on the same particles as the sample, full minus one controls for every gate that has to be drawn, and a viability dye. The control burden grows faster than the parameter count.

**How do I assign fluorochromes to markers?**

Brightest label on the dimmest antigen, dimmest label on the most abundant one, judged on your own instrument's lasers and filters rather than on a published index. Doing it the other way round is why sensible-looking panels resolve nothing.

**Why does my compensation change between runs?**

Usually a degraded tandem dye. The energy transfer link breaks down with light, fixation and freeze-thaw, and the degraded fraction emits in the donor's channel. Keep tandems dark and cold and compensate with the lot in use.

**Do I need a viability dye in every panel?**

Yes. Dead cells bind antibody non-specifically and autofluoresce, appearing as false positives in almost every channel. Use an amine-reactive dye where the cells will be fixed.

## Blotting equipment

More blots are lost at transfer than at detection. Large proteins move slowly and small ones pass through the membrane, the sandwich has to be assembled without bubbles, and the buffer composition changes what transfers. None of that is visible until the film is developed, which is why the transfer step deserves the attention usually given to the antibody.

1. **Match the transfer mode to the protein size**: Tank transfer is slow, gentle and best for very large proteins. Semi dry and cassette systems are fast and can drive small proteins through the membrane. Choose by the size range on the gel rather than by convenience.
2. **Verify transfer before probing**: A reversible total protein stain on the membrane, photographed, shows whether transfer worked and how evenly. It takes minutes and it prevents a day spent troubleshooting the antibody.
3. **Choose the gel chemistry for the range**: Low molecular weight proteins and peptides resolve poorly on a standard system and need a different buffer chemistry. Using the wrong gel puts the target in the dye front.
4. **Fix blocking against the target class**: Milk based blockers contain phosphoproteins and defeat phosphospecific antibodies. Protein free blockers suit those. The blocker is part of the method and a change is a method change.
5. **Use total protein for normalisation**: A stained membrane image gives a loading measure that does not depend on a housekeeping protein remaining constant under the treatment. It is now the expected normalisation for quantitative blots.
6. **Know what an alternative platform replaces**: Capillary and bead based systems automate the separation and detection and change the sample requirement and the dynamic range. They replace the labor, not the need to validate the antibody.

**Why is my large protein missing?**

It did not transfer. Very large proteins need longer, colder tank transfer, sometimes with methanol reduced and a low percentage gel. Stain the membrane and the gel afterwards to see where the protein went.

**Why is my small protein missing?**

It probably went through the membrane. Use a smaller pore size, shorten the transfer and add methanol, and confirm with a membrane stain.

**Should I normalise to a housekeeping protein?**

Total protein staining is safer, because housekeeping proteins change with treatment, confluence and stress more often than people assume. Show the total protein image with the blot.

**Are western blot kits worth it?**

For a standard workflow, they give consistency and remove preparation steps. For unusual targets or optimisation work, assembling the components gives the control that troubleshooting needs.

## Culture supplements

Medium is assembled from a base, supplements and additives, and in most laboratories only the base has a recorded lot. The supplements vary between lots, degrade at different rates and are frequently added by whoever is free, which makes them the largest uncontrolled variable in cell culture.

1. **Record every component and lot**: Base medium, serum or replacement, growth factors, antibiotics and any additive, each with a lot number, in the experimental record. Without it a change in behavior has no candidate explanation.
2. **Reserve supplement lots for a study**: Ask the supplier to hold a lot and buy enough to finish the work. A supplement lot change mid study is a variable that cannot be removed from the analysis afterwards.
3. **Prepare complete medium as a documented step**: Written composition, volume, date of preparation and a shelf life on the bottle. Growth factors and glutamine degrade, and an old bottle is a different medium.
4. **Match the vessel to gas exchange**: Shake flask fill volume, plate format and lid design decide oxygen transfer and evaporation. Edge effects in plates and oxygen limitation in over filled flasks are vessel problems, not biology.
5. **Keep stimulation reagents separate from supplements**: Reagents used to provoke a response are experimental treatments with their own controls, not medium components. Recording them in the same place as supplements confuses the two.
6. **Consider outsourcing where scale exceeds capability**: Expansion, banking and characterisation at scale are done better and cheaper by a service than by a research laboratory improvising, and the service returns documentation.

**Why did cells change behaviour after a cell culture media supplements change?**

Almost always a supplement lot rather than the base medium. Recording lots and reserving them for the length of a study is what makes this investigable and usually preventable.

**How long does complete medium last?**

Shorter than the base medium, because glutamine and growth factors degrade. Set a shelf life per formulation, label the bottle with the preparation date, and do not extend it informally.

**Do plate edge effects matter?**

Considerably. Evaporation from outer wells changes concentration and osmolality over a multi day experiment. Use inner wells for samples and fill the outer ring with buffer.

## Plates and labware

A plate is part of the measurement. Color changes crosstalk and background, surface treatment changes whether cells attach, material changes binding of hydrophobic compounds, and a seal that leaches changes the chemistry. Treating labware as interchangeable is the reason many assays drift when procurement changes supplier.

1. **Choose plate color for the readout**: Clear for absorbance and imaging, black to reduce crosstalk in fluorescence, white to maximize signal in luminescence. Using the wrong color costs sensitivity that no instrument setting recovers.
2. **Match surface treatment to the cells**: Untreated, tissue culture treated and specialized coatings behave completely differently for attachment and for compound binding. The treatment is a specification, not a default.
3. **Seal for the incubation you actually run**: Long incubations need a seal that holds without leaching and that can be removed cleanly. Test the seal with a plate of medium over the full incubation before adopting it.
4. **Treat filtration cut off as a specification**: Membrane material and pore size decide what passes and what binds. Protein binding to a filter is a real loss, and low binding materials exist for that reason.
5. **Standardize the set and record changes**: One plate, one seal, one tip family across the laboratory, with any substitution verified. Procurement changes are silent unless someone records them.

**Does plate colour really matter?**

Yes. Black plates reduce well to well crosstalk in fluorescence and white plates maximize luminescent signal. Running a luminescent assay in a clear plate loses a large part of the signal for no reason.

**Why do my outer wells read differently?**

Evaporation, usually, and sometimes uneven temperature. Fill the outer ring with buffer and use inner wells for samples in any incubation longer than a few hours.

**Can a seal on a PCR microplate affect the assay?**

It can. Some films leach plasticisers and some adhesives interfere with detection. Test a seal over the full incubation on a plate of medium before adopting it.

**Why do my edge wells always differ?**

Evaporation and thermal gradients affect the perimeter more, and the effect grows with well count and incubation time. Fill the perimeter with buffer, randomise treatments across the plate, and keep a plate identifier in the analysis.

**Does the dimensional standard make plates interchangeable?**

Only mechanically. It fixes footprint, height and well positions so plates fit instruments; well shape, working volume, surface treatment and optical properties are not fixed and are where plates actually differ.

## Nucleic acid recovery

Every recovery step loses material, and the loss is rarely measured. Gel extraction, difficult matrices, cross linked tissue and particle isolation each lose a different fraction for a different reason, and knowing the size of the loss once makes every subsequent low yield interpretable rather than mysterious.

1. **Measure recovery once per method**: Run a known quantity through the whole procedure and measure what comes out. The recovery figure turns future low yields into a comparison rather than a puzzle, and it takes one afternoon.
2. **Match lysis to what the sample defends itself with**: Cell walls, cross linking, inhibitors and nucleases each need a different countermeasure. Kits differ far more in this step than in the binding chemistry that follows.
3. **Handle cross linked material as its own problem**: Fixed and embedded material needs de-crosslinking and yields fragmented nucleic acid whatever is done. Design the downstream method around short fragments rather than hoping for long ones.
4. **Separate particle isolation from molecular extraction**: Isolating vesicles is a particle separation with its own contaminants and controls; extracting nucleic acid from them is a second step. Combining them in one description hides where the losses happen.
5. **Control the workspace for nucleases**: Nuclease contamination is a workspace property, not a sample property. Dedicated space, surface treatment and a periodic check are the controls, and a degraded preparation is usually a room problem.
6. **Use enzymatic clean up where a column cannot**: Residual single stranded or linear material that a column carries through can be removed enzymatically, which is faster and gentler than another purification round.

**How much do I lose in a DNA gel extraction?**

More than most people assume, and it varies with fragment size and gel percentage. Measuring it once with a known input tells you what to expect and whether a low yield is normal.

**Why does fixed tissue give short fragments?**

Because fixation cross links nucleic acid and protein and the reversal is incomplete and damaging. It is inherent to the material, so the library preparation and analysis have to be chosen for short input.

**How do I know my workspace is nuclease free?**

Run a sensitive RNA sample through the handling steps and check its integrity, and repeat periodically. Degradation that appears in the workspace and not in the stock is a room problem.

## Sequencing services

A sequencing project is decided before any sample is prepared, by the choice of what fraction of the nucleic acid is captured and how deeply it is read. Those two choices bound every question the data can answer, and neither can be changed afterwards without repeating the library preparation.

1. **Choose the library fraction from the question**: Selection for mature transcripts captures a defined fraction and discards the rest; depletion of ribosomal material keeps non coding and unpolyadenylated species. Degraded material usually forces depletion regardless of preference.
2. **Set depth from the effect size, not from a default**: Detecting a modest change in an abundant transcript needs far less depth than detecting a rare variant. Work the requirement out statistically before booking, because depth is the main cost lever.
3. **Match input requirements to what you have**: Every library chemistry has a minimum input and a preferred quality. Precious, degraded or low input samples need chemistries designed for them, and those have their own biases to declare.
4. **Agree the deliverables in writing**: Raw reads, alignments, the pipeline with versions and parameters, the reference used, per sample quality metrics and the code behind each figure. A report is a summary, not a deliverable.
5. **Randomise across preparation batches**: Batch, index set and lane all leave signatures. Distribute experimental groups across batches and hand the batch table to whoever analyses the data.
6. **Decide in house against a service honestly**: Owning a sequencer makes sense at sustained volume with a trained operator; below that, a service gives better data per unit cost and no maintenance. Synthesis capability follows the same logic.

**Total RNA seq or a poly A selected RNA library?**

Selection where mature transcripts are the question and the material is intact; depletion where non coding species matter or the material is degraded. They are not comparable, so mixing them within a study is a confound.

**How much depth do I need?**

It follows from the effect size and the abundance of what you are measuring. Calculating it beforehand is the difference between an adequately powered study and an expensive underpowered one.

**What should a provider hand over?**

Raw and aligned data, the exact pipeline with versions, the reference and annotation, per sample quality metrics and runnable code. Anything less cannot be reproduced or extended.

## CDMO selection

One word covers several different businesses. Making a bulk substance, filling it into containers under sterile conditions, handling a highly potent payload and synthesising a molecule chemically are separate capabilities with separate facilities and separate scarcity. Choosing a partner starts with naming which capability you actually need.

1. **Name the capability, not the category**: Bulk manufacture, sterile filling, potent handling and chemical synthesis are different plants. A partner excellent at one may subcontract another, which changes accountability and lead time.
2. **Treat sterile filling as the scarce step**: Filling capacity with the right container formats and the right containment is frequently harder to book than bulk capacity, and it sits at the end of the timeline where delay is most expensive.
3. **Assess the quality system, not the equipment list**: Deviation handling, change control, data integrity and the last inspection outcome predict batch success better than any equipment inventory. Ask to see the systems, not the suite.
4. **Establish who owns the process**: A platform partner runs their process and gives speed at the cost of portability. A contract manufacturer runs yours. Both are valid and the difference decides whether you can ever move.
5. **Confirm containment for potent payloads**: Highly potent material requires containment engineering, occupational exposure monitoring and cleaning verification between products. Not every biologics site can take that work, and it narrows the list early.
6. **Negotiate failure and reslot before signing**: What happens when a batch fails release, who pays, how quickly a slot reopens and what data you receive. These terms are far easier to agree before a campaign than during one.

**Why is sterile fill-finish harder to book than bulk?**

Because sterile filling suites are fewer, are shared across many products, and require format specific change over and validation. It is the step where timelines most often slip.

**What separates a good partner from a cheap one?**

The quality system and slot reliability. A failed batch or a missed slot costs far more than the difference in price, and both are predicted by deviation handling and inspection history rather than by equipment.

**Can a process be moved between partners?**

Only if you own it and it was documented to transfer. Platform processes generally are not portable, which is a reasonable trade to make knowingly and a painful one to discover.

## Small scale bioprocess

A small bioreactor is worth having only if what it tells you is true at scale. That requires it to be instrumented like the large vessel and operated at matched conditions on the parameter that actually limits the process, which is rarely volume and usually oxygen transfer, mixing time or shear.

1. **Decide what will be held constant across scales**: Power per volume, tip speed, oxygen transfer coefficient and mixing time cannot all be matched simultaneously. Choosing which one governs your process is the whole of scale up strategy.
2. **Instrument the small vessel like the large one**: Dissolved oxygen, pH, temperature and off gas at minimum, with the same control strategy. A vessel without the instrumentation of its target scale generates data that does not transfer.
3. **Run enough vessels to see variation**: A single small vessel gives one trajectory. Parallel vessels give the variability, which is what tells you whether a difference between conditions is real.
4. **Include the harvest in the model**: Clarification by depth filtration and concentration by tangential flow are where a good culture becomes an unusable harvest. Model them at small scale rather than discovering the problem at pilot.
5. **Record everything the large scale will need**: Feed strategy, timings, in process controls and their acceptance ranges, written as a process description rather than as a set of habits. That document is what transfers, not the vessel.

**Does a small vessel predict a large one?**

Only if it was designed to. Matching the governing parameter, instrumenting equivalently and running the same control strategy is what makes it a scale down model rather than a small experiment.

**How many parallel benchtop bioreactors are useful?**

Enough to estimate vessel to vessel variation, which usually means at least three per condition. A single vessel per condition cannot separate a treatment effect from run variability.

**Why include harvest at small scale?**

Because clarification and concentration frequently fail on cultures that looked excellent. Modeling them early prevents a pilot campaign discovering that the harvest cannot be processed.

## Service and calibration

Maintenance is bought reactively and costs most when it is. An instrument that fails mid batch costs the batch, the investigation and the schedule, and a calibration certificate that does not state its tolerances answers nothing. Both problems are solved by treating service as a scheduled program with a record rather than as an emergency.

1. **Set intervals by consequence, not by calendar**: An instrument whose failure stops production or invalidates results deserves a shorter interval and a spares holding. One whose failure is an inconvenience does not. Grade the estate and spend accordingly.
2. **Read what a certificate actually states**: A useful certificate names the method, the tolerances tested, the as found and as left values, the environmental conditions and the traceability of the reference. A pass with no numbers tells you nothing.
3. **Cost downtime, not parts**: Compare contracts on guaranteed response and parts availability rather than on annual fee. For a critical instrument, a slower contract is more expensive than a faster one.
4. **Decide in house against external per volume**: Where enough devices of one type exist, in house calibration with a traceable reference is cheaper and faster. Below that volume, an accredited external service is cheaper and more defensible.
5. **Keep one asset record with an owner**: Serial, location, service and calibration status, certificates and a named owner per instrument. Without it nobody knows what is due, and an audit finds out first.
6. **Plan for the failure that matters most**: Very low temperature storage is the failure with the largest consequence. Monitoring, alarms that reach a person, spare capacity and a rehearsed transfer are worth more than any contract.

**What makes a calibration certificate useful?**

Named method, stated tolerances, as found and as left values, environmental conditions and traceability of the reference standard. A certificate without as found values cannot tell you whether previous results were affected.

**Should balance calibration services be done in house?**

Where the volume of similar instruments justifies a traceable reference and a trained person, yes. Otherwise an accredited service is cheaper and produces a certificate that carries more weight.

**How should a service contract be compared?**

On response time, parts availability and what is excluded, rather than on the annual figure. Downtime on a critical instrument costs far more than the difference between contracts.

## Cold storage

A very low temperature freezer is a heat pump that rejects a large amount of heat into the room and draws power continuously, and its most important property is what happens when it fails. Capacity is the easiest thing to compare and the least important, because usable capacity is set by the rack system and by whether anything can be found.

1. **Survey the room before choosing the unit**: Heat rejection, power supply, floor loading and clearance for airflow and service. A unit installed in an unprepared room runs warm, fails early and warms everything around it.
2. **Compare on alarm and recovery behavior**: How quickly the unit recovers after a door opening, what it does on a power interruption, and how the alarm reaches a person. These decide whether a failure becomes an incident.
3. **Match the temperature to the material**: Not everything needs the lowest temperature, and colder storage is more expensive to run and to maintain. Match the class of material to the temperature it actually requires.
4. **Standardize the rack and box system**: Usable capacity is set by the rack, box and tube format, not by internal volume. One system across the laboratory makes racks interchangeable and positions describable.
5. **Write and rehearse the failure plan**: Spare capacity to transfer into, dry ice on hand, a call list and a rehearsed procedure. A plan nobody has practised is discovered to be incomplete during the failure.
6. **Split irreplaceable material across units and rooms**: Two locations with independent power and alarms is the cheapest insurance available for a collection that cannot be remade.

**What actually limits an ultra low lab freezer's capacity?**

The rack and box system and the index, not the internal volume. A standardized system with every position addressable gives far more usable capacity than a larger unit filled with mixed formats.

**Do ultra low temperature freezer manufacturers assume a dedicated room?**

They need a room that can take the heat and the power, with service clearance. Several units in an ordinary laboratory raise the ambient temperature enough to shorten the life of all of them.

**What should the alarm on an ultra low deep freezer do?**

Reach a person who can act, at any hour, with an escalation if unanswered. An alarm that sounds in an empty building at the weekend has no value.

## Assay kits

A kit is a validated recipe for one endpoint in one matrix, and its validity ends where those assumptions do. The recurring failures are using a kit outside its matrix, ignoring interference from the sample, and treating a proxy endpoint as the thing it stands in for.

1. **Match the endpoint to the question**: Metabolic activity, membrane integrity, cell number and strand breaks are different endpoints that behave differently. Choosing by convenience produces a number that answers a question nobody asked.
2. **Check the matrix the kit was validated in**: A kit validated in buffer or serum may not work in lysate, medium or a process intermediate. Ask what matrix the validation used and run a spike recovery in yours.
3. **Test for interference explicitly**: Colored compounds, high protein, chelators and inhibitors all interfere. A compound only control without cells, and a spiked sample, detect most of it in one plate.
4. **Understand the standard**: Every quantitative kit reports relative to its standard, and standards differ between suppliers. Switching kit changes the numbers, and comparing across kits needs a bridging experiment.
5. **Run scored assays blind and to a rule**: Assays scored by eye or by image analysis depend heavily on the rule and the scorer. Fix the rule, score blind, and report both.
6. **Keep schedule assays on a schedule**: Contamination screening is only useful before the results depend on it. Testing when something looks wrong means the affected experiments are already done.

**Why do two kits give different numbers?**

Different standards and different endpoints. Both can be internally correct, which is why a study should fix one kit and state it rather than mixing them.

**What does an endotoxin assay need before a sample?**

An inhibition and enhancement test in your matrix, showing that a spiked known amount is recovered. Without it a low result cannot be distinguished from a suppressed assay.

**Are cell proliferation assays measuring cells?**

Usually it measures metabolic activity, which tracks cell number only when metabolism per cell is constant. Compounds that affect metabolism directly break that assumption.

## Clinical lab informatics

When a sample comes from a person, the software has duties the sample tracker in a research laboratory never had: consent scope held against the material, custody recorded at every transfer, retention and destruction made deliberate, and every record attributable and unalterable. Those duties, not the feature list, decide which products are candidates.

1. **Start from the duties, not the features**: Consent scope, custody, retention and attributable records are obligations. List them first and use them to shortlist, because most products demonstrate convenience features and leave the duties to be discovered later.
2. **Map the custody chain on paper**: Every point where material changes hands, is split or is destroyed, and who records it. Points where nobody records anything are requirements, and they are where audits find problems.
3. **Decide what review has to be evidenced**: Batch record review and quality control approval have to be visible in the record with who did it and when. A system that allows review without recording it does not support the process.
4. **Check the audit trail by exporting one**: Ask to export an audit trail from a live system during evaluation. It is far more informative about a product's fitness than any demonstration of the interface.
5. **Separate laboratory from operational scope**: Trial monitoring, site support and recruitment are operational functions with their own systems. Trying to run them from a laboratory system produces a poor fit in both directions.
6. **Settle export before signing**: Format, whether the audit trail is included, whether attachments retain metadata, and whether it can be done without the supplier. It is the clause that matters most later and is negotiated least now.

**Is a biobank system just a sample tracker?**

No. It adds consent scope, chain of custody, retention and destruction control and attributable records. A research tracker with those bolted on is usually a poor fit, and the gap appears at the first audit.

**What does clinical laboratory automation actually change?**

Mostly the informatics: specimen routing, result capture, autoverification rules and the audit of those rules. The mechanical part is the visible half and the smaller problem.

**How is a batch record review evidenced?**

By a record showing who reviewed what and when, unalterable afterwards, tied to the batch or result. A signature on a printout is weaker and a review with no record is not a review.

## Cloning and delivery tools

Two products are sold as competent cells. One is optimized to take up plasmid efficiently, for cloning and library work. The other is optimized to express protein without degrading it or killing the host, which needs protease deficiency, tight repression and sometimes supplemented codon usage. Using the cloning strain to express is a common and quiet failure.

1. **Separate cloning strains from expression strains**: Cloning strains maximize uptake and plasmid stability. Expression strains minimize proteolysis and leaky expression. Their properties are close to opposite and the catalog names do not always make that obvious.
2. **Choose repression for a toxic product**: Where the product harms the host, leaky expression before induction selects for cells that have lost the construct. Strains and vectors offering tighter repression are the remedy, not lower induction.
3. **Match delivery to the cell, not to the protocol**: Chemical transformation suits laboratory strains, electroporation suits difficult ones and high efficiency needs, and suspension mammalian cells need formulations designed for them.
4. **Control arcing and cuvette hygiene**: Electroporation fails on salt carryover and reused cuvettes. Desalting the DNA and using fresh cuvettes removes most of the failures blamed on the instrument.
5. **Use transposon systems where a virus is unnecessary**: Transposon based integration gives stable expression without viral containment requirements, which simplifies approvals and handling for many applications.
6. **Design the workspace against amplicon carryover**: A dedicated setup enclosure with ultraviolet decontamination and a strict direction of travel controls contamination far better than any reagent. Automation helps by removing manual transfers, not by being fast.

**Why is my protein degraded?**

Frequently the strain. Expression strains deficient in major proteases exist for this reason, and switching strain is faster and cheaper than adding more inhibitors to the lysis buffer.

**Why do my transformants lose the plasmid?**

Leaky expression of a product the host does not tolerate. Tighter repression, a lower copy vector or a lower growth temperature before induction usually recovers it.

**When is electroporation worth the equipment?**

For difficult strains, library scale transformations where efficiency is the constraint, and primary or suspension cells. For routine cloning into laboratory strains, chemical transformation is simpler.

**Do I need high efficiency competent cells?**

For a library, a large construct or a scarce ligation, yes. For routine subcloning a standard chemical stock is ample, and a fraction of a vial will do. Aliquot so a vial is used once, because efficiency falls with each freeze and thaw.

**Why not clone and express in the same strain?**

Because the genotypes pull in opposite directions: a cloning strain protects the plasmid and transforms efficiently, an expression strain carries the polymerase the promoter needs and lacks degrading proteases. Build in one, express in the other.

**My protein is in inclusion bodies. What first?**

Lower the temperature after induction, reduce the inducer, and try a strain variant that suppresses pre-induction expression. Those are cheaper than changing host, which is the next step if they fail.

## Sterilisation equipment

An autoclave sterilises a load, not a chamber, and whether it succeeds depends on steam reaching every surface. Air trapped in a porous load or a wrapped instrument prevents that, which is why the cycle type and the way the load is packed matter more than the size of the machine.

1. **Match the cycle to the load type**: Gravity displacement suits liquids and unwrapped items where air can fall out. Porous loads, wrapped instruments and lumens need active air removal, and a gravity machine will not sterilise them reliably however long it runs.
2. **Pack the load so steam can reach everything**: Overloading, sealed containers and tightly wrapped packs all trap air. Packing is part of the validated process and should be written down and shown, not left to whoever is loading.
3. **Use indicators for what they prove**: Tape shows a pack has been through a process. A chemical indicator inside the pack shows conditions were reached at that point. A biological indicator shows the process kills spores. They answer three different questions.
4. **Validate with the load you actually run**: Cycle validation is performed on representative worst case loads with thermocouples or indicators at the hardest to reach points. Validating on an empty chamber proves nothing about a full one.
5. **Calibrate and record every cycle**: Temperature and pressure sensors drift, and a cycle printout or electronic record is what shows the load received the process. Keep the records against the load and the operator.
6. **Plan waste and liquid cycles separately**: Waste loads and liquid loads have different cycle requirements and different risks, including boil over. Separate cycle programs and separate containers are the usual answer.

**Does autoclave tape mean the load is sterile?**

No. It shows the pack was exposed to a process. Sterility assurance requires indicators placed where penetration is hardest, and periodic biological indicators, on a defined schedule.

**Gravity displacement autoclave or pre-vacuum?**

Gravity for liquids and unwrapped items. Pre-vacuum for porous loads, wrapped instruments and anything with a lumen, because active air removal is what lets steam reach the surfaces.

**Why did my load fail sterility?**

Almost always trapped air from packing or overloading rather than a machine fault. Repack to the validated arrangement and place an indicator at the point you suspect.

**Why does a liquid cycle take so much longer?**

Because the load cannot be vented quickly: liquid above its boiling point would erupt. The cycle cools slowly under controlled pressure, vessels must be loosely capped or vented, and the fill level has to leave headspace.

## IHC reagents

An immunohistochemistry result is produced by five layers acting together, and a failure in any of them looks like a failure of the antibody. Retrieval decides whether the epitope is available, blocking decides the background, the detection layer decides the sensitivity, and the chromogen decides what can be seen alongside it.

1. **Optimize retrieval per antibody**: Buffer, pH, temperature and duration each change epitope availability, and the optimum differs between reagents even on the same tissue. Optimize on tissue known to be positive and record the conditions with the protocol.
2. **Block three things, not one**: Endogenous peroxidase, endogenous biotin where a biotin system is used, and non specific protein binding are separate problems needing separate steps. Skipping any of them produces a characteristic background.
3. **Prefer a polymer detection system where biotin is a problem**: Polymer based detection removes the endogenous biotin question entirely and usually gives cleaner backgrounds in biotin rich tissues such as liver and kidney.
4. **Choose the chromogen for the context**: Chromogens differ in color, in permanence and in whether they can be combined for multiplexing. A chromogen that fades or dissolves in the mounting medium ruins an archive.
5. **Include the controls on the same slide where possible**: A positive tissue, a negative tissue and a no primary control. Internal positive elements within the section are better still, because they experienced identical processing.
6. **Fix the whole protocol before comparing sections**: Fixation time, section thickness, retrieval, incubation times and temperature all change the result. Comparing sections processed differently compares processing.

**Why is my stain weak, and is it the primary antibody?**

Retrieval, most often. In cross linked tissue the epitope has to be recovered, and the conditions differ between reagents. Optimize retrieval before increasing antibody concentration.

**Where does high background come from?**

Endogenous enzyme activity, endogenous biotin in a biotin based system, or insufficient protein blocking. Each has a distinct appearance and a distinct remedy, and adding more blocking protein does not fix the first two.

**What controls does a section need?**

A known positive, a known negative and a no primary control, ideally processed in the same run. Internal positive elements within the section are the strongest control because they saw identical conditions.

**Chromogenic or fluorescent?**

Chromogenic for architecture, archival stability and bright-field reading; fluorescent for multiplexing and quantification. The question, not the fashion, decides.

**Does an increased pathway marker mean more activity?**

Not on its own. Markers on structures that are continuously formed and consumed report a balance, so an increase can mean more formation or less clearance. A blocking step is what separates them.

## Method validation and E&L

A number from an unvalidated method is an observation. Validation is what turns it into evidence, and it is scaled to the decision the number supports: early development needs a method shown to be fit for purpose, and a filing needs the full exercise. Container studies follow the same logic from the other direction.

1. **Scale validation to the phase**: Early work needs demonstrated specificity, a working range and repeatability. A filing needs accuracy, intermediate precision, robustness, solution stability and a documented protocol executed as written. Doing the full exercise too early wastes effort; doing it too late delays the filing.
2. **Generate the degradants before validating specificity**: A stability indicating method has to separate the product from what it turns into. Forced degradation under acid, base, oxidation, heat and light produces those species so the method can be shown to resolve them.
3. **Separate extractables from leachables**: Extractables are what a container can give up under deliberate stress; leachables are what it actually gives up into your product over its shelf life. The first defines what to look for and the second is what has to be controlled.
4. **Choose standards with traceability**: Certified reference materials, and system suitability standards run every sequence, are what make a result comparable to anyone else's. Their availability for your analyte is worth checking before the method is designed around them.
5. **Write the protocol before running it**: Acceptance criteria decided in advance are what make a validation a validation. Criteria set after the data exist describe the data rather than testing the method.
6. **Plan transfer at the same time**: A method that cannot be run elsewhere is a bottleneck. Write it with enough detail for another laboratory, and validate the transfer as a defined exercise rather than as an informal handover.

**How much analytical method validation does early work need?**

Enough to show the method is fit for the decision it supports: specificity, a working range and repeatability, documented. The full exercise belongs later, and doing it early is a common waste.

**Why run forced degradation?**

To generate the degradation products a stability indicating method has to separate. Without them, specificity cannot be demonstrated against the species that actually matter.

**Is leachables and extractables testing the same study as an extractables leachables screen?**

No. Extractables are generated by deliberately stressing the container material; leachables are measured in the actual product over its shelf life. The first informs the second and does not replace it.

## Digital PCR

Partitioning a reaction into thousands of small compartments turns amplification into a counting experiment, which removes the standard curve and tolerates inhibitors better than a conventional reaction. It also costs considerably more per sample, so the platform earns its place on the questions that need it rather than on everything.

1. **Identify the question that needs absolute counting**: Copy number, rare variant detection at low fraction, and quantification without a reference are where partitioning wins. Routine relative expression is not, and running it there is expensive.
2. **Cost per data point, not per plate**: Reagents, partitioning consumables and the lower number of samples per run together set the real cost. Compare against the conventional platform on the same experiment, not on list prices.
3. **Understand what partition count buys**: More partitions extend the upper dynamic range and improve precision at low copy number. The specification that matters is partitions per reaction and their volume reproducibility, not the headline throughput.
4. **Optimize the assay before partitioning**: An assay with poor separation between positive and negative amplitude will not be reliably classified into partitions. Optimizing annealing temperature and primer concentration first is what makes the counting work.
5. **Plan the dilution**: Partitioned counting saturates when almost every partition is positive. Samples usually need dilution into the countable range, and finding that range is part of the method.

**When is a digital PCR system worth the cost?**

For copy number, rare variant detection and absolute quantification without a reference material. For routine relative expression across many samples, conventional real time amplification remains far cheaper per data point.

**Does it really remove the standard curve?**

Yes, because the readout is a count of positive partitions rather than a comparison to a curve. That is its main advantage, along with better tolerance of some inhibitors.

**Why is my partition classification poor?**

Usually assay optimisation. Poor amplitude separation between positive and negative partitions is an assay problem, and it has to be fixed before the instrument can classify reliably.

## Mass spectrometry services

Mass spectrometry is the most capable and the most demanding instrumentation in a laboratory. Owning it means owning source cleaning, vacuum maintenance and a person who understands both the chemistry and the software; buying the answer means finding a laboratory whose accredited scope covers your method and whose matrix experience matches yours.

1. **Count the samples and the methods**: A steady flow of one or two established methods justifies owning. Occasional, varied and difficult work is usually better bought, because the expertise rather than the instrument is what you are short of.
2. **Name the matrix in every conversation**: Ion suppression, source contamination and carryover are matrix problems. A laboratory or a vendor with experience of your matrix is worth more than a better specification without it.
3. **Choose the analyser from the question**: Targeted quantitation of known compounds points to a triple quadrupole. Identification, screening and retrospective analysis point to high resolution. Elemental work is a different instrument class entirely.
4. **Budget the running cost honestly**: Gases, vacuum pump service, source cleaning, a service contract and the operator. On these instruments the recurring cost is a large fraction of ownership and it is rarely in the business case.
5. **Inspect a used instrument on the parts that wear**: Vacuum system condition and hours, source state, detector age and whether the control software runs on a supported operating system. The last one ends more used purchases than the hardware does.
6. **Check accredited scope when buying a service**: Accreditation covers named methods. Confirm yours appears on the published schedule, and ask what is subcontracted.

**Triple quad mass spec or high resolution?**

Targeted quantitation of known analytes at volume points to the first; identification, unknown screening and the ability to reinterrogate data point to the second. Many laboratories eventually need both.

**What is the real running cost of a mass spectrometry laboratory?**

Gas, vacuum servicing, source cleaning, consumables and a service contract, plus the operator's time. It is substantial and it is the part most often omitted from a purchase justification.

**Are used instruments from mass spectrometry companies sensible?**

It can be, from a serviced fleet with records. The risks are vacuum system condition, detector age and unsupported control software, and all three are checkable before purchase.

## Common cell lines

The most used cell lines are used because they grow reliably, and that reliability is bought with genomic instability and loss of differentiated function. Each is a good model of something specific and a poor model of the tissue it came from, and knowing which is which is the difference between a defensible experiment and a convenient one.

1. **State what the line models**: Write down which property of the biology the line is standing in for. If that property is one the line is known to have lost, the experiment needs a different system or an explicit caveat.
2. **Authenticate and screen on arrival**: Short tandem repeat profiling against a reference and a contamination screen, before the line joins the culture area. The most used lines are also the most misidentified, precisely because they are shared so widely.
3. **Fix a passage window**: These lines drift with passage, and the drift is faster in some than others. Bank early at low passage, work from the bank, and record passage number with every experiment.
4. **Control the differentiation state**: Several of these lines behave very differently depending on confluence, medium and whether they have been differentiated. Fix the protocol and describe the state, because the same line can be two systems.
5. **Pair with a primary system where the claim needs it**: Where a conclusion depends on physiology the line has lost, a primary or stem cell derived system is what supports it. The line is then the screening tool and the primary system the confirmation.

**Are HepG2 cells a model of liver?**

Partly. It retains some hepatic functions and has lost much of the drug metabolising capacity of primary hepatocytes, so it models some questions well and metabolism poorly. Say which you are relying on.

**Why authenticate a line everybody uses?**

Because widely shared lines are the most likely to have been swapped somewhere in the chain, and morphology will not reveal it. Profiling is inexpensive and settles it permanently for that stock.

**Do these lines need a passage limit?**

Yes. They drift, and a study spanning a wide passage range has an uncontrolled variable. Bank at low passage and return to the bank rather than extending a culture indefinitely.

## Centrifuges and sample prep

Centrifugation is specified by the rotor and the sample, not by the maximum speed on the front panel. Whether a rotor exists for your tube, whether the temperature holds under load, and whether the sample survives the shear are the questions, and homogenisation adds the same trade between disruption and destruction.

1. **Choose the rotor before the instrument**: The rotor determines which tubes fit, what force is achievable and how the sample bands. An instrument with an impressive maximum and no rotor for your format is unusable.
2. **Confirm the temperature holds under load**: Refrigerated units warm as they work, particularly at high speed with a full rotor. Ask for the temperature achieved at your intended speed and load rather than the setpoint range.
3. **Match disruption to the molecule**: Bead beating, rotor stator homogenisation and sonication each disrupt effectively and each shear long nucleic acid and damage labile proteins. Choose the gentlest method that opens the sample.
4. **Use spin filters where a pellet is not the point**: Concentration and buffer exchange in a centrifugal filter is faster and gentler than precipitation, and the membrane cut off and binding characteristics are the specification that matters.
5. **Treat extraction as part of the analysis**: Solid phase extraction cleanliness determines instrument uptime downstream. Where a mass spectrometer or a chromatograph is fed, preparation is where the reliability comes from.
6. **Balance, inspect and log**: Rotor fatigue is a real hazard. Log rotor usage, inspect for corrosion and pitting, and retire on the manufacturer's schedule rather than on appearance.

**What actually limits a centrifuge?**

The rotor: which tubes it takes, what force it reaches and its remaining service life. Two instruments with the same headline speed differ entirely in what they can spin.

**Why did my sample warm up?**

Refrigerated units heat under load, and the setpoint is not the sample temperature at speed. Ask for the achieved temperature at your speed and load, and pre cool the rotor.

**Bead beating or rotor stator?**

Bead beating for tough tissue and cell walls, rotor stator for soft tissue and larger volumes. Both shear long nucleic acid, so a gentler method is preferable when fragment length matters.

**Does a refrigerated instrument keep my sample cold?**

Only if it controls temperature under load. Some instruments merely pre-cool and then do not heat, and a fast rotor warms itself considerably. Ask for the temperature achieved at your speed with your rotor fitted.

## Containment and sterilisation

Containment equipment answers one question first: who or what is being protected. A fume hood protects the operator by pulling air away; an isolator protects the product by keeping air in. Choosing the wrong one protects nobody properly, and both impose demands on the building that are cheap to plan and expensive to retrofit.

1. **Decide who is protected**: Operator protection means air moves inward and is extracted. Product protection means filtered air moves over the work. Protecting both requires specific designs, and assuming a hood does both is a common and serious error.
2. **Size for the apparatus and its extract load**: A large enclosure moves a large volume of conditioned air, and that cost is continuous. Sizing to the actual apparatus rather than to a maximum saves money every day the building runs.
3. **Treat sash position as a control**: Face velocity depends on how far the sash is open, and the safe working position is a specification. Sash alarms and stops exist because the control depends on the user.
4. **Match sterilisation to the load**: Steam sterilisation with active air removal is required for porous loads and wrapped items. Liquids, waste and instruments need different cycles, and the load arrangement is part of the validated process.
5. **Design the waste route with the equipment**: Where waste is autoclaved before disposal, the bins, the transport and the cycle are one system. Deciding it afterwards produces a route people work around.
6. **Involve the building early**: Extract capacity, make up air, floor loading and service access are building constraints. A hood that cannot be ducted, or that unbalances the ventilation, is an expensive discovery.

**Does a fume hood protect the sample?**

No. It protects the operator by drawing air inward, which pulls room air across the work. Product protection needs a cabinet or an isolator with filtered air moving over the sample.

**When is an isolator justified?**

When the product must be protected to a defined standard, when the material is highly potent, or when the operation cannot be interrupted by room conditions. It is a facility decision rather than an equipment purchase.

**Does the fume hood sash position really matter?**

Yes. Face velocity and therefore containment depend on it, which is why hoods have marked working heights and alarms. Working with the sash high is the commonest containment failure.

## Food and environmental testing

For environmental and food work the sample, not the analysis, usually decides the result. Where it was taken, how many increments, how it was preserved and how long it took to reach the laboratory bound what any method can tell you, and a perfect analysis of an unrepresentative sample is a precise wrong answer.

1. **Design the sampling plan first**: Number of increments, location, depth or point of collection, and how they are composited. For heterogeneous material this dominates the uncertainty and it is decided before anyone contacts a laboratory.
2. **Preserve and transport to the method's requirement**: Container type, temperature, preservative and holding time are specified by the method. Exceeding a holding time invalidates the result regardless of how well the analysis was run.
3. **Name the method and check the scope**: Different standards give different numbers for the same property. State the method and confirm it appears on the laboratory's published accreditation schedule.
4. **Ask about matrix recovery**: Detection limits assume a recovery from a matrix. Ask for recovery data on a matrix like yours rather than accepting a figure generated on a clean one.
5. **Establish chain of custody where it matters**: For forensic, regulatory and dispute related work the custody record is part of the evidence. Agree the documentation before the first sample is taken.
6. **Specify the report contents**: Method, sample description and condition on receipt, holding times, result with uncertainty, any deviations and a signature. Agree it in advance rather than discovering an omission when it is needed.

**Why do soil testing labs' results vary so much between samples?**

Heterogeneity, almost always. Soil, food and environmental matrices vary over short distances, and a single grab sample can differ substantially from the average. A proper increment plan is the remedy.

**Does holding time really invalidate a result?**

For many analytes, yes, because they degrade or transform. Methods specify holding times for that reason, and exceeding one makes the number unusable for a regulatory purpose.

**What should I check about a local laboratory?**

That its accreditation schedule includes your method and matrix, what it subcontracts, and its turnaround to a signed report. Proximity matters mainly because it shortens holding time exposure.

## Blotting reagents

Most of what determines a blot happens before the antibody. The lysis buffer decides which proteins are in the sample at all, the sample buffer decides how they run, the membrane decides how much binds and stays bound, and the software decides what the number means. Each is chosen by default far more often than by reason.

1. **Choose lysis for the compartment**: A mild buffer leaves membrane, nuclear and cytoskeletal proteins in the pellet. A strong denaturing buffer solubilises almost everything and destroys native interactions. Decide which your target and your question need, and blot the pellet once to check.
2. **Keep inhibitors in the buffer, not beside it**: Protease and phosphatase inhibitors have to be present before the cells meet the buffer, made fresh, and dated. Adding them afterwards is the commonest reason a signal is lost.
3. **Do not confuse sample buffer with running buffer**: The sample buffer denatures, reduces and loads; the running buffer carries the current and sets the ion front. Substituting one for the other is a mistake that produces a characteristic mess.
4. **Match the membrane to the protein and the detection**: Nitrocellulose gives low background and is fragile; polyvinylidene fluoride binds more and tolerates stripping and reprobing. For small proteins the pore size matters more than the material.
5. **Choose the blocker against the target class**: Milk based blockers carry phosphoproteins and defeat phosphospecific detection. Protein free and albumin based blockers suit those cases. A blocker change is a method change.
6. **Keep the raw image and the settings**: Any number from a blot depends on the exposure and the analysis regions. Software that exports only an adjusted image makes the number indefensible.

**Why is my target missing from the RIPA buffer western blot lysate?**

It is probably in the pellet. Membrane, nuclear and cytoskeletal proteins need stronger extraction than a mild buffer provides. Blot the pellet once and the question is settled for that cell type.

**Milk or albumin blocking in TBS western blot buffers?**

Albumin or a protein free blocker for phosphospecific detection, because milk contains phosphoproteins. Milk is cheaper and effective otherwise. The target decides.

**Which western blot membrane: nitrocellulose membrane for western blot or polyvinylidene fluoride?**

Nitrocellulose for low background and easy handling; the fluoride membrane for higher binding capacity, durability and reprobing. Pore size matters more than material for very small proteins.

## Bioreactors and bioprocessing

A bioreactor is bought as hardware and used as a control system. What separates units is not the vessel but the sensors, the control strategy and whether the data leaves in a form anyone else can use, and the same is true of the partners who operate them at scale.

1. **Judge a vessel on its control system**: Dissolved oxygen, pH, temperature, agitation and gas mixing under closed loop control, with off gas where possible. A vessel with fewer measured variables produces data that does not transfer to a controlled process.
2. **Check how data leaves**: Continuous logging in an open format, exportable without the supplier, is what lets a run be analyzed and compared. A controller whose data lives only in its own software is a future problem.
3. **Match microbial and mammalian requirements separately**: Microbial fermentation demands oxygen transfer and heat removal that a mammalian vessel is not built for. A unit sold as suitable for both is a compromise and the compromise should be understood.
4. **Treat skids and continuous operation as facility decisions**: Continuous or connected processing changes buffer demand, footprint, control and how a batch is defined. It is a plant decision rather than an equipment upgrade.
5. **Divide the work explicitly with a partner**: Who owns the process, who develops the analytics, who holds the raw data and what transfers at the end. Ambiguity here is the commonest source of dispute at the end of a campaign.
6. **Ask what the partner's constraint is**: Every site has a scarce step, whether suite time, analytics or release testing. The honest ones will tell you, and it predicts your timeline better than a capacity figure.

**What separates one laboratory bioreactor from another?**

The control system and the data path, not the vessel. Sensors, closed loop control and exportable logging are what make a run comparable and transferable.

**Can one vessel serve microbial and mammalian work?**

With compromises. Microbial cultures demand far higher oxygen transfer and heat removal, and a unit sized for mammalian work will limit them. Where both are routine, two vessels usually beat one compromise.

**What should be agreed before booking a slot?**

Process ownership, the analytical package and who develops it, raw data access, what happens if a batch fails release, and the reslot terms. These are far easier to agree before a campaign than during one.

## Perturbation tools

Silencing and cutting answer the same question differently. Silencing is partial, reversible and fast; cutting is complete, permanent and slower, and each has an off target profile of its own. At screen scale the decision is dominated by delivery and by how much material the format consumes.

1. **Choose knockdown or knockout from the biology**: Essential genes and dose response questions suit partial, reversible silencing. Clean loss of function suits cutting. Neither is generally better and the choice should be stated with a reason.
2. **Solve delivery before scaling**: In a screen, delivery efficiency multiplies through every well. Establish it with a fluorescent control and a known positive perturbation before committing library scale material.
3. **Size the screen on representation**: Library scale work sets a minimum number of cells per reagent that has to be maintained at every step. Calculate it backwards from library size, because a screen below representation is noise.
4. **Use an inducible system where timing matters**: Where a perturbation is lethal or where the timing of onset is the question, an inducible system separates the perturbation from the selection. It adds a construct and answers a question a constitutive system cannot.
5. **Budget the plasmid**: Screens consume more DNA than plans allow for. Large scale preparation, or a preparation service, belongs in the plan rather than discovered mid screen.
6. **Keep sample preparation enzymes separate in the design**: Proteases, glycosidases and nucleases used to prepare samples are analytical tools. Recording them alongside perturbation reagents confuses two different parts of the experiment.

**Silencing or cutting for a screen?**

Silencing for essential genes, partial loss and fast, reversible perturbation. Cutting for clean loss of function and for a selection based readout over longer timescales. Both are used and the choice follows the readout.

**What makes a screen fail?**

Representation, almost always: too few cells per reagent at transduction, selection, sorting or sequencing. The remedy is arithmetic done before the screen rather than analysis afterwards.

**Why include a fluorescent delivery control?**

Because it separates a failed perturbation from a failed delivery. Without it, a null screen result cannot be interpreted at all.

## Sequencing services and QC

Amplicon surveys and whole genome sequencing answer different questions about the same organisms. A marker gene survey gives composition cheaply at the price of resolution and functional information; whole genome sequencing gives both and costs more per sample in sequencing and considerably more in analysis.

1. **Choose amplicon or genome from the question**: Composition across many samples suits a marker gene survey. Strain level resolution, function and antimicrobial genes need whole genome sequencing. Deciding after sampling is the expensive route.
2. **Name the region for an amplicon survey**: Different variable regions give different apparent composition and different resolution, and results across regions are not comparable. Fix the region and report it with every result.
3. **Run contamination controls in low biomass work**: Reagents and kits carry their own bacterial DNA, which dominates low biomass samples. Extraction blanks and no template controls carried all the way through sequencing are essential rather than optional.
4. **Quantify libraries properly before pooling**: A pool balanced on a rough concentration gives wildly uneven read counts. Quantification by amplification or by an appropriate fluorometric method, plus a size check, is what makes a pool even.
5. **Check adapter and index compatibility**: Index collisions and adapter mismatches waste a run. Confirm the index set is compatible with the instrument's chemistry and with any other libraries sharing the flow cell.
6. **Agree deliverables and controls in writing**: Raw reads, the pipeline with versions, the reference database and its version, per sample quality metrics, and the blanks sequenced alongside. A composition table alone cannot be re-analyzed.

**Does a 16S rRNA sequencing service tell you what the community does?**

No. It tells you what is there, at a resolution set by the region chosen. Function requires whole genome or transcriptome sequencing, and inferring it from composition is an estimate rather than a measurement.

**Why sequence blanks?**

Because extraction kits and reagents carry bacterial DNA that can dominate a low biomass sample. Without blanks carried through the whole process, contaminant taxa are indistinguishable from real ones.

**Why are my read counts so uneven?**

Library quantification before pooling, usually. Rough concentration measurements do not account for adapter ligation efficiency or fragment size, and the pool ends up unbalanced.

## Culture consumables and markers

A cultured population drifts toward whatever the plastic, the medium and the passage schedule select for, and the markers used to identify it change with that drift. Consumables that look neutral, serum, coatings and plate surfaces, are frequently the strongest selective pressure in the experiment.

1. **Choose markers that survive culture**: Several identity markers are lost or gained within a few passages in culture. Pick at least one marker known to be stable in your system, and pair it with a functional check where one exists.
2. **Treat serum as a variable, not a supplement**: Serum composition varies by lot and carries vesicles, growth factors and hormones. Where those matter, depleted or defined alternatives exist and change other properties too, which has to be characterised rather than assumed.
3. **Use surface chemistry deliberately**: Attachment, low attachment and coated surfaces select different populations and induce different behavior. A plate change is a biological change and should be recorded as one.
4. **Keep blocking reagents out of the detection layer**: Purified immunoglobulin used for blocking or as an isotype reference is a control reagent. Recording it with detection antibodies confuses two roles that need separating in the protocol.
5. **Record passage, lot and surface with every result**: Passage number, serum lot, coating and plate type. Those four fields explain most unexplained differences between experiments in the same laboratory.

**Why did my population lose its marker?**

Culture selects, and many identity markers are state dependent. Reduced staining can mean drift rather than loss of cells, which is why a stable marker plus a functional check is better than one marker.

**What does depleted serum change?**

It removes vesicles and, depending on the method, some proteins and lipids with them. Cells frequently behave differently in it, so the change has to be characterised rather than treated as neutral.

**Do low attachment plates change biology?**

That is their purpose. Preventing attachment selects for anchorage independent behavior and induces different signalling, which is a deliberate intervention rather than a neutral vessel choice.

**Is one endothelial marker enough?**

Not across beds. Arterial, venous, lymphatic and microvascular endothelium differ, and lymphatic vessels are routinely missed by a general marker. State the bed and include a bed-specific marker where it matters.

## Metabolite and contamination kits

Small kits are bought to answer a question in an afternoon, and they do that well inside their assumptions. The recurring problems are reading a proxy as the thing itself, running an assay in a matrix it was not validated in, and treating a threshold assay as a quantitative one.

1. **Separate metabolite from damage readouts**: A metabolite concentration reports consumption or production. A released intracellular enzyme reports membrane damage. Both are measured in the same supernatant and they answer different questions.
2. **Know whether the endotoxin format gives a number**: Gel clot methods give a pass or fail against a threshold; kinetic and chromogenic formats give a concentration. Choosing the wrong one for the decision produces a result that cannot be used.
3. **Run inhibition and enhancement before any sample**: A spiked recovery in your own matrix is what shows the assay can see endotoxin in that sample. Without it a low result cannot be distinguished from a suppressed assay.
4. **Keep contamination screening on a schedule**: Screening after a surprising result means the affected experiments are already done. New arrivals in quarantine and continuous cultures on a fixed interval is the pattern that works.
5. **Check the matrix and the standard**: Every kit is validated in a matrix against a standard. Serum, medium and lysate interfere differently, and a dilution series plus a spiked recovery detects most of it in one plate.

**Does an LDH assay kit measure toxicity?**

It measures membrane damage, which is one late consequence of toxicity. Cells that stop dividing without lysing show nothing, which is why it is paired with a metabolic or count based readout.

**Which endotoxin format should I use?**

A threshold method where a pass or fail against a limit is the decision, and a kinetic or chromogenic method where a number is needed. Both require inhibition and enhancement testing in your matrix.

**How often should mycoplasma be tested?**

On arrival in quarantine, and on a fixed interval for cultures in continuous use. Testing only when something looks wrong means the contaminated cultures have already produced data.

**Will a kit work in my sample?**

Only test it. Run a spiked recovery across the working range in your own matrix; a standard curve in buffer says nothing about interference from reducing agents, color, turbidity, chelators or a competing enzyme.

**Can I use a stored standard curve?**

It is the most common shortcut and the usual explanation for drifting values. Run the curve on every plate, include a known control sample and record the fit.

## Chromatography media and solvents

Solvent grade is a specification, not a preference: trace ultraviolet absorbing impurities set the baseline, and a gradient concentrates them into peaks that look like analytes. The stationary phase question is separate and mechanical, and mixing analytical and preparative media in one conversation is how purification trains get designed badly.

1. **Buy solvent by grade, and match it to detection**: Gradient grade solvents are specified for low ultraviolet absorbance, and for mass detection the requirement is low involatile residue. Using a lower grade produces a rising baseline and ghost peaks that are frequently blamed on the column.
2. **Filter and degas, and change the filters**: Inlet filters and mobile phase filtration protect pumps and columns and clog quietly. Replacing them on a schedule prevents a pressure problem being diagnosed as a column failure.
3. **Condition bare silica properly**: Normal phase separations on unbonded silica are sensitive to water content in the mobile phase and need consistent conditioning. Retention that drifts across a sequence is usually this.
4. **Control the column temperature**: Retention and selectivity change with temperature, and ambient laboratory temperature varies more than people assume. A column heater removes a source of drift for a modest cost.
5. **Choose preparative media by capacity and mechanism**: Affinity, hydroxyapatite and ion exchange media are chosen by what they bind and how much, and they behave differently on scale up and on cleaning. Analytical column reasoning does not transfer.
6. **Plan cleaning and reuse for resins**: A preparative resin is reused, and its cleaning regime and lifetime are part of the cost. Ask for cycle life data under a cleaning protocol like the one you intend to run.

**Do HPLC solvent grades really matter?**

Yes, particularly with gradients and ultraviolet detection, because trace impurities accumulate on the column at low organic content and elute as peaks. For mass detection, involatile residue matters more than absorbance.

**Why does my retention drift across a sequence?**

Temperature or, on bare silica, mobile phase water content. Both are controllable, and a column heater plus consistent solvent handling removes most of it.

**Can analytical media be scaled up?**

Not directly. Preparative media are chosen for capacity, pressure tolerance and cleanability, and the economics are dominated by resin lifetime rather than by resolution.

**Do HPLC solvent bottle caps matter?**

They do, for two reasons that have nothing to do with the cap itself. An unsealed bottle lets the organic evaporate, so a premixed mobile phase changes composition over a long sequence, and an open aqueous bottle grows something within days. A capped bottle with a filtered vent keeps the composition and keeps the dust out.

**Can acetone HPLC grade be used as a wash?**

As a wash for a column being stored or for tubing, sometimes, but it absorbs strongly in the low ultraviolet, so any left in the system shows up as a baseline that will not settle. It is also a poor choice anywhere near a mass detector. Where the intent is to remove strongly retained material, a stronger alcohol or a defined regeneration gradient is the safer answer.

## Benchtop analysers

Small analysers are bought because they are affordable and are then relied upon for decisions. What separates a usable instrument from a decorative one is whether it runs quality control material, whether its calibration is traceable, and whether the result carries the method with it.

1. **Ask what decision the number supports**: A trend indication and a release decision need very different levels of control. Deciding this first determines whether quality control material, traceable calibration and a records requirement apply.
2. **Confirm quality control material exists**: An analyser without appropriate control material cannot demonstrate it is working. Availability and cost of that material is a running cost and occasionally the reason an instrument is impractical.
3. **Match path length and volume to the sample**: Micro volume instruments trade path length for sample size and behave differently on turbid or highly absorbing samples. A cuvette instrument is more forgiving and needs more material.
4. **Choose absorbance, color or fluorescence deliberately**: A colorimeter reads defined wavelengths for a defined chemistry; a spectrophotometer scans; a fluorimeter reports emission and is far more sensitive for suitable analytes. They are not substitutes.
5. **Cost service, space and waste**: Counting, tablet handling and materials testing instruments each need space, service and a waste or disposal route. Those costs are frequently larger than the instrument for the smaller units.
6. **Keep the method with the result**: Instrument, method, calibration date and control result alongside the number. Without them the value cannot be compared to one taken next month.

**Colorimeter vs spectrophotometer for a benchtop analyser?**

A colorimeter measures at fixed wavelengths for defined chemistries and is cheap and robust. A spectrophotometer scans and suits method development and unknown samples. Choose by whether the wavelengths are fixed.

**Are micro volume instruments accurate?**

Within their assumptions, yes, and they are sensitive to turbidity, bubbles and surface tension because the path length is formed by the drop. For difficult samples a cuvette instrument is more reliable.

**What makes a small analyser usable for decisions?**

Traceable calibration, available quality control material run on a schedule, and a record that ties the result to the method and the calibration state. Without those it is an indicator rather than a measurement.

**What actually decides the cost of a clinical analyser?**

The reagent relationship. Most are closed systems, so price the instrument over its life at your own test volume including reagents, controls, calibrators and service; the instrument is often the smallest line.

**Can a research instrument report a clinical result?**

No. Labeling is a property of the product: research use only means it may not be used to report a clinical result, whatever the laboratory's own accreditation.

## Microscope cameras and optics

A camera cannot add information the optics did not deliver, and most microscope camera purchases are made on sensor specifications that the objective makes irrelevant. Matching pixel size to the optical resolution, and field of view to what has to be seen, is the whole of the decision.

1. **Match pixel size to the optical resolution**: Sampling finer than the optics resolve adds file size and no information; sampling coarser throws resolution away. The objective and the coupling determine the useful pixel size, and the camera is chosen to match.
2. **Choose sensor size for the field you need**: A larger sensor sees more of the field the objective projects, at the cost of price and sometimes of vignetting. Decide the required field before comparing sensors.
3. **Weigh sensitivity against speed**: Cooled, larger pixel cameras suit dim fluorescence and long exposures; fast, small pixel cameras suit brightfield and live imaging. One camera rarely does both well.
4. **Insist on parfocality and parcentricity for measurement**: Objectives that stay in focus and centred when switched are what make multi magnification work practical, and are essential for any measuring application.
5. **Calibrate every objective and camera combination**: A traceable stage graticule per combination, repeated on a schedule and recorded. Without it, a measurement from an image is an estimate.
6. **Inspect used equipment on the optics and the software**: Fungus and haze in objectives, sensor condition, and whether the camera software runs on a supported operating system. The last of these ends more used purchases than anything mechanical.

**Does more resolution mean a better image?**

Only up to the optical limit. Beyond it, extra pixels add file size and noise without information. Matching the sensor to the objective is what improves images.

**One camera for fluorescence and brightfield?**

Possible and usually a compromise. Dim fluorescence wants sensitivity and cooling; brightfield and live work want speed and color. Two cameras on one stand is common for that reason.

**What is needed before measuring from an image?**

A calibration against a traceable graticule for that exact objective, coupling and camera, recorded and repeated on a schedule. Any change in the light path invalidates the previous calibration.

## GC instruments

A gas chromatograph is three decisions: how the sample gets in, what separates it, and what notices it coming out. Instruments are compared on the last of those and limited by the first, because inlet discrimination and gas quality decide what the detector ever sees.

1. **Choose detectors by what they respond to**: A universal detector responds to almost everything and is insensitive. Element selective detectors ignore the matrix and see only their element. Mass detection identifies as well as quantifies. The analyte and the matrix choose, not the budget.
2. **Specify the inlet for the sample**: Split, splitless, on column and headspace introduce sample very differently and discriminate differently against high boiling components. A method that loses heavy analytes is usually losing them at the inlet.
3. **Fix gas purity and protect it**: Carrier gas purity determines column life and baseline stability, and traps are cheap next to a column. Specify the grade, install traps and put their replacement on the maintenance schedule.
4. **Match oven performance to the method**: Ramp rate and cool down time set the cycle time, which is what determines throughput. Two instruments with identical separations can differ substantially in samples per day.
5. **Calibrate against traceable standards**: Peak area becomes concentration only through a calibration with standards of known content, bracketed by check standards in every sequence. That, not the instrument, is what makes the number defensible.

**Which of the gas chromatography detectors should a general laboratory have?**

A universal detector for broad screening plus mass detection where identity matters. Element selective detectors are added for named applications rather than for completeness.

**Why do heavy components disappear?**

Inlet discrimination, usually. Split injection favors volatiles unless the inlet temperature and liner are right, and the effect looks like a sample problem.

**Does adding mass detection change the laboratory?**

Yes. It brings a vacuum system, source maintenance, tighter sample cleanliness requirements and a different skill set. It is a workflow change rather than a detector swap.

**What does a gas chromatography laboratory need beyond the instrument?**

Gas of a stated purity with traps and a regulator per line, a bench that is not shared with anything that vibrates or smells of solvent, extraction for the oven exhaust, and somewhere to keep standards cold and dark. The gas supply is the part underestimated most often: a cylinder change is a change in the method unless the purity is specified.

**Is gas chromatography testing worth outsourcing?**

For an occasional sample set, yes, because the instrument time is the smaller part of the cost and the method development is the larger one. It stops being worth it once the same method runs weekly, at which point the laboratory is paying a service to maintain expertise it should hold itself.

## Plate pipetting

A full plate head turns ninety six transfers into one and turns every plate into a consumable decision. Tips are consumed a plate at a time, the head has to reach the bottom of the wells you actually use, and any tip that does not seal is ninety six wells wrong rather than one.

1. **Check head reach against your plates**: Deep wells, skirted plates and low volume formats each impose a geometry. A head that cannot reach the bottom of a deep well, or that fouls a skirt, is unusable for that format however good it is elsewhere.
2. **Cost tips per plate, not per box**: A full plate head consumes a rack of tips per transfer. At the intended run rate that recurring cost frequently exceeds the depreciation on the device, and second sourcing may be restricted.
3. **Match the device to how full the plates are**: Partial plate work is better served by an eight or twelve channel device; a full head used on a half plate wastes tips and time. Most laboratories need both.
4. **Use a dispenser where the reagent is common**: Filling a plate with one reagent is a dispenser's job, not a pipette's. It is faster, uses a tube rather than tips, and is more consistent.
5. **Verify plate devices as a set**: A multichannel device has to be verified channel by channel, because one weak channel produces a column of outliers that looks like a biological effect. Software that records per channel results is worth having.

**Is a full plate head worth it?**

When whole plates are transferred repeatedly, yes, because it removes ninety six chances to make an error. For partial plates and varied work it is expensive and inflexible.

**Why does one column of my plate read differently?**

Frequently a single weak channel on a multichannel device. Verify per channel rather than as an average, and rotate the plate once as a diagnostic.

**Dispenser or pipette for filling a plate?**

A dispenser, whenever the same reagent goes into every well. It is faster, more consistent and does not consume tips.

**Eight channels or a whole plate head?**

A row device first, in almost every case: it is cheaper, more flexible and tolerant of plate geometry. A whole-plate head earns its place on throughput, once the plates per week justify the tip cost and the reach constraints.

## Cytometry fluorophores

Fluorophore brightness is a property of the dye, the laser that excites it and the filter that collects it, which means brightness rankings only apply to a particular instrument. Assigning the brightest available dye to the dimmest antigen is the single most effective panel design decision, and it cannot be made from a catalog alone.

1. **Rank brightness on your own configuration**: A dye that is bright on one laser is dim on another. Rank the conjugates you can actually excite and detect, using beads if necessary, and keep that ranking with the panel designs.
2. **Put the brightest dye on the dimmest antigen**: Resolution is set by the product of dye brightness and antigen density. Lineage markers expressed at high density can sit on modest dyes; a low density activation marker cannot.
3. **Block before staining any myeloid population**: Fc receptors bind conjugates through the constant region and produce signal in every channel. This is a required step, and the commonest cause of an uninterpretable myeloid stain.
4. **Account for spread, not just spillover**: Compensation removes the average spillover and leaves the variance, which widens negative populations in the receiving channel. Place rare and dim measurements where spread is lowest.
5. **Record the panel completely**: Marker, clone, fluorophore, supplier, lot, titre and detector for every reagent. Without those seven fields the panel cannot be rebuilt, and a discrepancy cannot be investigated.

**Which fluorophore is brightest?**

It depends on the laser and filter configuration. Some conjugates are consistently near the top on common configurations, and the ranking still has to be confirmed on the instrument you use.

**Why do I see CD14 flow cytometry signal in every channel?**

Unblocked Fc receptors, in most cases. It produces plausible positive signal rather than obvious noise, which is what makes it dangerous, and a blocking step removes it.

**Does sorting change the panel?**

It adds constraints: the cells have to survive, the sort gates have to be resolvable at speed, and viability and aggregate exclusion become essential rather than optional.

## Imaging modalities

Modality is chosen by the sample, not by prestige. A thin, dim specimen is imaged best by collecting all the light there is; a thick one needs the out of focus light rejected, either optically or computationally. Choosing by what sounds most advanced produces long acquisitions of bleached samples.

1. **Start from thickness and brightness**: Thin and dim favors widefield with a sensitive camera, because it collects every photon. Thick favors optical sectioning. That single question settles most of the decision.
2. **Consider deconvolution before buying sectioning**: Computational deconvolution of a widefield stack recovers much of what sectioning would have rejected, at no cost in photons. For light sensitive samples it is frequently the better route.
3. **Match contrast method to the specimen**: Unstained, transparent and reflective samples need contrast rather than resolution: phase, differential interference, darkfield or reflected illumination. No amount of camera improves a specimen with no contrast.
4. **Specify working distance for manipulation**: Where the sample has to be reached, dissected or injected, working distance and stand geometry matter more than numerical aperture.
5. **Isolate the stand where it needs it**: Vibration limits resolution and long acquisitions more than most people expect. An isolation table is cheap next to the instrument and is frequently omitted.

**Is confocal always better than wide field microscopy?**

No. Confocal rejects out of focus light, which is essential in thick samples and wasteful in thin ones, where widefield with a sensitive camera collects far more signal per unit of illumination.

**What does deconvolution actually do?**

It reassigns blurred light computationally using a model of the optics, which sharpens a widefield stack. It works well on well sampled data with a good point spread function and cannot invent information.

**When is a darkfield microscope useful?**

For unstained specimens and small scattering objects, where contrast rather than resolution is the limitation. It is a contrast technique and not a resolution improvement.

## Sterility and release testing

The classical sterility test takes fourteen days, which is longer than some products exist. Rapid methods shorten that and have to be shown equivalent or better before they can be used, which is a validation exercise rather than a purchase, and it sits alongside the rest of a release package.

1. **Validate a rapid method against the compendial one**: Equivalence has to be demonstrated for the organisms and the product in question, including a method suitability test showing the product does not inhibit growth. Adopting a rapid method without that is not a shortcut, it is an unvalidated result.
2. **Run method suitability before any sample**: Products with antimicrobial properties inhibit the test itself. Demonstrating that a known low inoculum still grows in the presence of the product is what makes a negative result meaningful.
3. **Control the upstream sterilisation**: Sterility at release depends on validated sterilisation cycles and aseptic handling upstream. A test is a check on a process, not a substitute for one.
4. **Include stability and identity in the package**: Genetic stability for a cell or vector product, identity, purity and potency alongside sterility. A release package with sterility alone answers one question out of several.
5. **Test incoming materials to a written plan**: Raw material testing on receipt, to a defined plan with acceptance criteria, prevents a failure that only becomes visible at release when the batch is already made.
6. **Decide what happens on a positive result**: Investigation route, retest policy where permitted, and the disposition of the batch. Deciding this in advance is what keeps a positive from becoming an argument.

**Can a rapid method replace the compendial test?**

With a validation demonstrating equivalence or superiority for your product and the relevant organisms, and with the regulator informed as required. Without it the rapid result is supporting information rather than a release test.

**Why does method suitability come before a rapid sterility test?**

Because a product with antimicrobial activity can suppress growth in the test, turning a contaminated batch into a passing result. Demonstrating recovery of a low inoculum in the presence of the product is what excludes that.

**What else belongs in a release package?**

Identity, purity, potency, endotoxin and, for cell and vector products, stability and adventitious agent testing. Sterility is one line in a panel rather than the panel.

## Clean air enclosures

These two enclosures look similar and do opposite things. One blows filtered air outward over the work to protect the product; the other draws air inward to protect the person. Using a clean air bench for hazardous material blows the hazard at the operator, which is why the distinction is a safety matter rather than a preference.

1. **Answer who is protected before anything else**: Product only means a clean air bench. Operator only means a fume hood. Both means a biological safety cabinet of the appropriate class, or an isolator. Nothing else about the specification matters until this is settled.
2. **Read a classification as a property of the air**: A cleanliness class describes particle counts in the working zone under defined conditions. It says nothing about containment, and a very clean bench can be entirely unsuitable for hazardous work.
3. **Choose materials for the chemistry**: Where strong acids or solvents are used, the cabinet material and the extract path have to tolerate them. A stainless enclosure and a polypropylene one are chosen for different chemistries.
4. **Understand what a filtered enclosure replaces**: A recirculating filtered enclosure removes particulates and, with the right media, some vapours. It does not remove everything, and the filter has a life that has to be monitored rather than assumed.
5. **Certify on a schedule and act on the trend**: Face velocity, containment and filter integrity are tested annually. A result that has been marginal for successive tests is telling you something about the system rather than about the test.

**Can I use a clean air bench for cell culture with a pathogen?**

No. It protects the culture and blows air at the operator. Work with an agent requiring containment belongs in a biological safety cabinet of the appropriate class.

**What does an air cleanliness class mean?**

A limit on particle counts in the working zone under stated conditions. It is a statement about cleanliness, not about containment, and the two are frequently conflated in purchasing.

**Is a ductless enclosure a fume hood substitute?**

Only within limits. It filters rather than exhausts, the filter media have to match the chemistry, and the filter has a finite capacity that must be monitored. For many chemistries it is not an adequate substitute.

**Which enclosure protects the operator from an aerosol?**

A biological safety cabinet. A chemical enclosure protects against vapour and not against an aerosol released inside a recirculating system, and a clean-air bench blows the aerosol toward the operator.

## Nuclear and senescence antibodies

Nuclear proteins are hard for structural reasons: they are tightly bound, frequently scarce, and several of them move between compartments in response to exactly the treatments being studied. A whole cell lysate and a standard fixation are the wrong starting points for most of them.

1. **Extract the nucleus rather than the cell**: Chromatin and envelope associated proteins stay in the pellet with a standard buffer. A nuclear extraction, with the pellet checked, is what puts them on a blot.
2. **Prefer activity where abundance is very low**: For enzymes present at a handful of copies per cell, an activity assay is more sensitive and more meaningful than an attempt to detect the protein directly.
3. **Name the isoform for envelope proteins**: Splice isoforms of nuclear envelope proteins differ in tissue distribution and in disease relevance, and reagents differ in which they see. The datasheet has to state it.
4. **Treat RNA binding proteins as mobile**: These proteins relocalise into granules and between compartments on stress. A change in staining pattern is frequently relocalisation rather than expression, and fractionation or imaging is what distinguishes them.
5. **Denature before detecting incorporated nucleotide**: An antibody against an incorporated base cannot reach it in double stranded DNA. Acid or enzymatic denaturation is a required step, and it damages other epitopes in the same sample.

**Why can I not detect telomerase?**

Because it is present at extremely low copy number in most cells. Activity assays are the standard approach, and a detection attempt without enrichment is likely to fail whatever the reagent.

**Does a stress granule marker measure stress?**

It measures granule formation, which is one response to some stresses. Total protein does not change, so the measurement is the number and size of granules per cell, which is an imaging readout.

**Why does BrdU antibodies detection need denaturation?**

The epitope is inside double stranded DNA and inaccessible. Denaturation exposes it and damages protein epitopes at the same time, which is why co-staining needs a compatible protocol or a different label chemistry.

## Inflammation enzyme antibodies

An enzyme that is nearly absent at rest and strongly induced is easy to detect and easy to over-interpret. The band appears, the fold change looks enormous, and the number depends almost entirely on the exposure and the loading control. Reading these reagents well is mostly a matter of quantitation discipline.

1. **Establish the linear range before quantifying**: A target that goes from nearly absent to abundant will saturate the detection long before the untreated lane is visible. Run a dilution series and choose an exposure where both ends are on scale.
2. **Normalise to total protein**: Housekeeping proteins change under the stimuli that induce these enzymes. Total protein staining is the safer normalisation and should be shown alongside.
3. **Prepare microsomes for membrane bound enzymes**: Drug metabolising enzymes sit in the endoplasmic reticulum, and a whole cell lysate dilutes them. A microsomal preparation concentrates them and is the conventional sample.
4. **Measure secreted proteins in the medium**: A secreted lectin or cytokine leaves the cell, so a lysate reports what has been made and not released. Measure both and say which is being reported.
5. **Use a compartment marker as a control, not a result**: Structural organelle markers are references for fractionation and imaging. Treating a change in one as a biological finding usually reflects a change in the preparation.
6. **Pair induction with activity where the claim is functional**: An induced enzyme is not necessarily an active one. Where the conclusion is about activity, an activity assay or a product measurement is what supports it.

**Why does my induced enzyme look infinitely induced?**

Because the untreated lane is below detection at the exposure used. A dilution series and a longer exposure of the control, or a more sensitive substrate, put both ends on scale so a real fold change can be reported.

**Do I need microsomes?**

For membrane bound drug metabolising enzymes, yes. They are concentrated in a microsomal fraction and diluted in a whole cell lysate, which is why the conventional sample is the fraction.

**Lysate or medium for a secreted protein beside an iNOS antibody?**

Medium for what was released, lysate for what has been made and retained. They are different quantities and reporting one as the other is a common error.

## Model and syngeneic lines

A model line is chosen for the question it makes answerable, not for how well it represents a tumour. Syngeneic lines exist so that immunology can be studied in an animal with a working immune system, and they are mouse biology; human lines answer human target questions and need an immunodeficient host.

1. **Match the line to the host strain**: A syngeneic line only grows without rejection in the strain it came from. Using it in another background produces a rejection experiment, which is occasionally the point and usually not.
2. **Decide whether the question is target or immune**: Human target pharmacology usually needs a human line and an immunodeficient host, which removes the immune system from the model. Immune questions need a competent host and therefore mouse biology. Very few models give both.
3. **Use a secreted reporter for longitudinal work**: A reporter released into blood or medium allows repeated measurement without sacrificing the animal or the culture, which reduces numbers and increases statistical power.
4. **Characterise what the line releases if that is the subject**: Vesicles, cytokines and metabolites released by a line are part of its phenotype and vary with confluence and medium. Characterise them under the conditions of the experiment rather than from the literature.
5. **Treat array screens as hypothesis generators**: A multiplex array indicates which analytes changed and at what approximate magnitude. Confirming with a quantitative assay for the ones that matter is what turns a screen into a result.
6. **Document the line as you would a reagent**: Source, passage, authentication, screening and the exact growth conditions. Model results that cannot be reproduced usually differ in one of those five.

**Why use a syngeneic line?**

Because it grows in an immunocompetent animal of the matching strain, which is the only way to study an immune response to a tumour in vivo. The cost is that the biology is mouse rather than human.

**Can a human line be used in an immunocompetent host?**

No, it is rejected. Human lines require an immunodeficient or humanised host, and that choice removes or reshapes exactly the immune biology a syngeneic model preserves.

**Are antibody arrays quantitative?**

Semi quantitative at best. They rank and indicate magnitude across many analytes, and anything a conclusion depends on should be confirmed with a validated single analyte assay.

## Purification systems

Purification platforms differ mainly in pressure and in how fractions are decided. Flash systems run fast at low pressure with disposable cartridges and suit synthetic chemistry; preparative liquid chromatography runs at pressure with reusable columns and finer control. Extraction avoids the column entirely when the chemistry allows.

1. **Size from the loading, not the volume**: How much can be loaded before resolution degrades sets the throughput. Measure it on an analytical column and scale by stationary phase mass rather than by column dimensions alone.
2. **Decide disposable cartridge against packed column**: Cartridges remove packing variability and cleaning, and cost more per run. Reusable columns are cheaper per run and require packing skill and a cleaning regime. Volume decides.
3. **Pack properly or buy packed**: A poorly packed bed channels, and the separation never recovers. Frits, packing technique and bed compression determine whether a self packed column performs, and for critical work a packed column from a supplier is the safer economy.
4. **Set fraction collection on a real trigger**: Threshold, slope or time, chosen against the actual chromatogram, with an analytical method ready to check the fractions. Collecting blind and analyzing later wastes material.
5. **Consider extraction before chromatography**: Where the target and the impurity differ in partition behavior, liquid or solid phase extraction is far cheaper and faster than a column. Try it first for simple separations.

**Flash or preparative liquid chromatography system?**

Flash for fast, lower resolution separations of synthetic mixtures at low pressure. Preparative liquid chromatography where resolution, reproducibility and detection control matter and the material justifies the cost.

**Are disposable cartridges worth it?**

For varied, occasional separations, yes, because they remove packing and cleaning variability entirely. At sustained volume the consumable cost favors a reusable packed column.

**Why does my self packed column channel?**

Uneven packing, bed compression or a damaged frit. All three produce a separation that cannot be recovered by method changes, and repacking is the remedy.

## Protein characterisation

A protein preparation is described by a concentration far more often than by what is in it. Aggregates, truncated species, host proteins and residues from the purification all travel with the target, and each is invisible to some methods and obvious to others, which is why characterisation uses a pair.

1. **Use two orthogonal purity methods**: A denaturing gel and a size exclusion separation see different problems: the gel finds truncations and contaminants by size, the column finds aggregates and multimers. Either alone gives an optimistic answer.
2. **Confirm identity independently**: A band at the right size is consistent with the protein and does not identify it. Mass spectrometry, either intact mass or peptide mapping, is what confirms it, and it is worth doing once per construct.
3. **Measure concentration by a stated method**: Absorbance with a calculated extinction coefficient, a colorimetric assay against a standard, or amino acid analysis give different numbers. Choose one, state it, and use it consistently.
4. **Measure the residues that matter downstream**: Imidazole, reducing agents, detergents, endotoxin and residual protease each interfere with something. Measure the ones your assay cannot tolerate rather than all of them.
5. **Account for filtration losses**: Protein adsorbs to membranes, and a sterile filtration step can remove a meaningful fraction of a dilute preparation. Low binding membranes exist and the loss should be measured once.

**Is a single band on a gel enough?**

No. A denaturing gel is blind to aggregates and to species of identical size, and it is a low resolution method. Pairing it with a size exclusion separation catches most of what it misses.

**How should protein concentration be measured?**

By a method suited to the protein and stated with the result. Absorbance is accurate for proteins with aromatic residues; colorimetric assays are matrix and standard dependent; amino acid analysis is the reference.

**Why does my dilute protein disappear on filtration?**

Adsorption to the membrane. It is substantial for dilute solutions, and low binding membranes plus a measured recovery are the practical response.

**Which protein analyser do I need?**

It depends which number you want: crude protein content in a commodity, concentration in a preparation, or the distribution of species in a purified product. Those are three different instrument families sharing one name.

## Vector production services

Making vector in house consumes containment space, plasmid and a person, and for occasional needs a service is faster and cheaper. What a service delivers varies enormously, and the difference is in the titre method, the purity and whether the paperwork lets you use the material for what you intended.

1. **Name the titre method in the order**: Physical particle, genome and functional titres differ by large factors, and only a functional titre on a cell type like yours predicts what you will get. Specify it, and specify the cell line it is measured on.
2. **Specify purity for the target cells**: Crude supernatant is adequate for robust lines and inadequate for primary cells, where residual medium components and cellular debris cause toxicity. Concentrated and purified preparations exist for that reason.
3. **Ask what capsid content is reported for adeno associated vector**: Full to empty ratio changes the effective dose and the immune load. A titre with no capsid content figure describes half the product.
4. **Confirm containment and shipping before ordering**: Vector arrives under containment requirements and institutional approval, and the paperwork has to exist before the box does. This is the commonest cause of a delayed start.
5. **Order the controls with the vector**: An empty or non targeting construct made in the same run is the control that matters, because it shares the preparation and its contaminants. Ordering it later gives a control from a different batch.
6. **Agree what happens on a low titre batch**: Titre varies between productions. Whether a low batch is remade, credited or accepted with an adjusted dose should be settled in the order rather than after it.

**Should I make or buy lentivirus?**

Buy for occasional needs, for primary cell work needing purified material, and where containment space is scarce. Make when production is frequent enough to justify the containment, the plasmid and the person.

**Which titre matters?**

The functional titre on a cell type resembling yours. Physical particle counts are much higher and do not predict transduction, and comparing suppliers on unnamed titres compares nothing.

**Why order a control construct in the same batch?**

Because it shares the production, the purification and any contaminants. A control from another batch controls for the construct and not for the preparation.

**Why is my transduction efficiency low?**

Usually the cell type or the cassette. Measure a functional titre on your own cells rather than relying on one from a standard line, and check the insert size, since packaging falls as the genome approaches its limit.

**In house or buy packaged vector?**

Buy it unless you produce regularly: a service delivers a titred preparation faster than a laboratory learns the process. Compare on functional titre, volume, whether concentration is included and the quality control.

## Serum and raw materials

Serum is a biological product with a country of origin, a testing history and lot to lot variation that exceeds most experimental effects. Qualifying a lot means testing it on your cells for the thing you care about, not reading a certificate, and the same logic applies to every other biologically derived material on the bench.

1. **Read the certificate for origin and testing**: Country of origin, adventitious agent testing and any irradiation are the regulatory facts. They tell you whether a lot may be used and nothing about whether it will work for you.
2. **Test the lot on your own cells**: Growth rate, plating efficiency and any assay endpoint that matters, compared against the current lot. This is the qualification, and it takes a week that saves a study.
3. **Reserve enough of a good lot**: Ask the supplier to hold it and buy the study's requirement in one purchase. Changing lot mid study introduces a variable that cannot be removed afterwards.
4. **Treat dissociation and buffer reagents as variables**: Enzyme activity, chelator concentration and even plate surface change how cells detach and recover. Fix them, record the lots and treat a change as a protocol change.
5. **Write a specification a supplier can be held to**: Named tests, acceptance ranges, documentation required with each lot and what happens on a failure. Without it a complaint has nothing to rest on.
6. **Record the handling equipment too**: Homogenisation method and counting chamber affect yield and the number that everything is normalised to. They belong in the method rather than in the drawer.

**Does a serum certificate mean the lot will work?**

No. It documents origin and agent testing. Performance in your system is not covered by it, which is why lot testing on your own cells is the standard practice.

**How much serum should be reserved?**

Enough to complete the study, with a margin. A lot change mid study is a confound that cannot be corrected afterwards, and suppliers will hold a lot on request.

**Do dissociation reagents matter?**

Considerably. Enzyme activity varies between lots and over storage, and over digestion damages surface antigens and viability. Fix the lot, the concentration and the time.

**Should serum be heat inactivated?**

Usually not. It destroys complement and some growth factors with it, and most modern culture does not need it. Where it is needed, supplier-inactivated serum is more consistent than a water bath.

**Why does vesicle work need depleted serum?**

Because ordinary serum carries its own vesicles in numbers that swamp what the cells release. Depletion also removes other components, so an adaptation period and a growth check belong in the method.

**Which vesicle isolation method is best?**

None in general: centrifugation co-isolates aggregates, size exclusion is cleaner at lower yield, precipitation is fast and dirty, affinity capture selects a subpopulation. State the method and confirm important conclusions with a second one.

## Incoming quality control

Testing everything on receipt is expensive and testing nothing is indefensible. A workable program tests identity on everything, tests performance on the few materials that determine outcomes, and relies on a qualified supplier and a certificate for the rest, with the reasoning written down.

1. **Test identity on everything**: Confirming a container holds what the label says is the irreducible minimum and it catches the errors that matter most. It is usually a fast instrumental check rather than a full analysis.
2. **Rank materials by consequence**: Which materials, if wrong, would invalidate a study or a batch? Those get performance testing. The rest are accepted on certificate from a qualified supplier, and the ranking is written down.
3. **Qualify the supplier rather than every lot**: An audited supplier with a stable process and a meaningful certificate reduces the testing burden legitimately. Supplier qualification is the work that makes reduced testing defensible.
4. **Use a service laboratory for rare tests**: A test needed twice a year does not justify an instrument, a method and a trained analyst. Outsourcing it is cheaper and produces a report from an accredited scope.
5. **Investigate failures with the right technique**: Failure analysis on a material, a device or an electronic assembly uses different methods and delivers an evidenced opinion. Agree the scope and the techniques before the investigation starts.
6. **Keep the supporting systems calibrated**: Ovens, balances and thermometers used in incoming testing need their own calibration, and preventive maintenance on the equipment is part of the same assurance chain.

**Do I have to test every incoming material?**

Identity, yes, as a minimum. Full testing of everything is rarely proportionate; a documented risk ranking with supplier qualification is the standard and defensible approach.

**When should testing be outsourced?**

When a test is needed rarely, requires accreditation, or needs an instrument the laboratory cannot justify. The report also comes from a defined accredited scope, which is worth something.

**What makes a certificate of analysis trustworthy?**

A qualified supplier, named methods, actual results rather than conforms statements, and periodic verification of a lot against your own testing. A certificate alone from an unqualified supplier is a claim.

## Plate handling automation

Automating plate work pays in a predictable order: dispensing first, then washing, then sealing and stacking, then scheduling. Buying the scheduler first, which is the usual instinct, produces an expensive cell that nobody can reprogram and that still waits for a person to fill the plates.

1. **Measure the manual workflow first**: Time each step and count the interruptions. The step that consumes the most hands on time is the one to automate, and it is frequently not the one people complain about.
2. **Start with dispensing**: Filling plates with a common reagent is the highest volume, lowest complexity step and the one a dispenser does better than a person. It is usually the cheapest real improvement available.
3. **Cost dead volume and cassettes**: A dispenser holds a minimum volume in its tubing that is lost every run, and cassettes are consumables with a life. For expensive reagents this can outweigh the time saved.
4. **Add washing where the assay is wash sensitive**: Plate washing is where immunoassay variability is created, and a well maintained washer removes operator differences. It is the second purchase for a plate based laboratory.
5. **Buy a scheduler only once the devices exist**: A scheduling layer coordinates instruments; it does not replace them. Buying it before the devices, or before drivers exist for them, produces a cell that cannot be assembled.
6. **Confirm driver support for used equipment**: Second hand liquid handlers are frequently good value mechanically and orphaned in software. Confirm a supported driver and a control computer that can still be maintained.

**What should be automated first, and where do laboratory automation companies fit?**

Dispensing, in almost every plate based laboratory, because it is high volume and low complexity. Washing is second where the assay depends on it. Scheduling is last.

**What is dead volume and why does it matter?**

The reagent held in the tubing and manifold that cannot be dispensed and is lost each run. For a cheap buffer it is irrelevant; for an expensive reagent it can dominate the running cost.

**Is used automation worth buying?**

Mechanically often yes, and the risk is software: unsupported drivers, an obsolete control computer and no path to integration. Confirm both before buying.

## Modified oligo chemistry

Modified nucleotides change the rules that oligonucleotide design tools were written for. A chemistry that raises duplex stability lets a probe be much shorter, which improves discrimination and breaks the melting temperature calculations most software applies by default.

1. **Recalculate melting behavior for the chemistry**: High affinity modifications raise duplex stability substantially, so a conventional calculation overestimates the length required. Use the supplier's tool for that chemistry rather than a generic one.
2. **Place modifications where discrimination matters**: Modified bases positioned over the discriminating position give the largest gain in specificity. Scattering them along a probe raises affinity without improving discrimination.
3. **Characterise a modified sequence properly**: Mass confirms the composition and not the position of a modification. Where position matters, ask for the analytical evidence rather than assuming synthesis went as ordered.
4. **Plan the purification a conjugate needs**: Coupling an oligonucleotide to a peptide, a dye or a protein produces a mixture that has to be separated, and the yield is frequently modest. Budget the purification and the characterisation.
5. **Think twice about in house synthesis**: A synthesiser brings reagents, waste, maintenance and a trained user. It pays where turnaround is critical and volume is sustained, and rarely otherwise.
6. **Use design tools as filters**: Software excludes obviously poor designs and cannot predict which good one works. Order several and test, particularly with modified chemistry where prediction is weaker.

**Why can a modified probe be so short?**

Because the modification raises duplex stability, so fewer bases reach the same melting temperature. A shorter probe discriminates single base differences better, which is the main reason to use the chemistry.

**Does mass spectrometry confirm a modification?**

It confirms the total composition. Position within the sequence usually needs a separate analysis, and for a critical probe it is worth asking the supplier what evidence they hold.

**Is an in house synthesiser worth it?**

For groups making sequences continuously with tight turnaround, yes. For occasional orders the reagents, maintenance and trained operator cost far more than buying.

**When are LNA probes worth the cost?**

For short targets such as small regulatory RNAs, for single-base discrimination and for hybridisation in situ where a short probe penetrates better. For ordinary amplification against a long target an unmodified design is cheaper and works.

## Laboratory refrigeration

A laboratory refrigerator is judged on uniformity across the chamber, recovery after the door opens and whether an alarm reaches a person. Domestic appliances cycle their temperature deliberately, which is fine for food and unacceptable for reagents and samples with a stability claim.

1. **Specify uniformity, not setpoint**: A chamber whose shelves differ by several degrees stores some samples outside their condition. Ask for mapping data, and map your own unit once when it is installed and loaded.
2. **Reject domestic appliances for stored material**: They control by cycling around a setpoint and have no alarm or record. For anything with a stability claim or a replacement cost, that is not a saving.
3. **Plan heat and power for very low temperature units**: They reject substantial heat and draw power continuously. Several in one room raise the ambient temperature and shorten the life of all of them.
4. **Monitor independently of the unit**: A logger that records to a separate system survives the failure of the unit and produces the history you will be asked for. The unit's own display is not a record.
5. **Calibrate ovens and incubators the same way**: Heated equipment used for a process step needs uniformity mapping and calibration on a schedule, with the reference traceable.
6. **Write and rehearse the transfer plan**: Spare capacity, dry ice, a call list and a rehearsed procedure. A plan nobody has practised is discovered to be incomplete during the failure.

**Why not use a domestic fridge instead of a laboratory refrigerator freezer combo?**

Because it cycles its temperature by design, has no alarm and produces no record. For reagents with a stated storage condition, or samples that cannot be replaced, none of that is acceptable.

**What does mapping involve?**

Loggers placed through the loaded chamber over a period covering normal use, showing the range actually experienced. It takes a week and tells you which shelves can be used for what.

**Should monitoring be part of the unit?**

Independent monitoring is better, because it survives a controller failure and keeps a record the unit does not. Many units offer both and the independent record is the one that answers questions.

## Centrifuge consumables

The tube is rated, and the rating is the limit. A vessel rated below the force it is spun at deforms or fails, and a failure at speed is a serious event rather than a lost sample. Matching vessel, adaptor and rotor is the whole of safe centrifugation.

1. **Check the vessel rating against the run**: Every tube and bottle carries a maximum force. Exceeding it deforms plastic and can shatter glass, and the rating depends on temperature and on whether the vessel is full.
2. **Match adaptors, and balance carefully**: Adaptors are part of the rotor assembly and have their own limits. Balance by mass rather than by eye, and balance the adaptors as well as the tubes.
3. **Check chemical compatibility**: Solvents craze and weaken plastics, sometimes without visible change before the spin. Where solvents are involved, use a material rated for them and inspect before reuse.
4. **Use spin columns where separation, not pelleting, is wanted**: Buffer exchange and desalting in a spin column is faster and gentler than precipitation, and the resin bed and volume range are the specification that matters.
5. **Understand what a vacuum concentrator does**: It removes solvent under reduced pressure with centrifugal force preventing bumping. It is a drying instrument rather than a separator and it has its own temperature limits.
6. **Log rotor use and retire on schedule**: Rotors accumulate fatigue with each run at speed. A usage log and a retirement date, enforced, is the control, and inspection alone is not sufficient.

**Can I spin any tube in any rotor, table top centrifuge or high speed?**

No. Tubes carry a maximum force rating, and adaptors have their own. Exceeding either risks deformation or failure, and a failure at high speed is a safety event rather than a lost sample.

**Do solvents damage centrifuge tubes?**

Several do, sometimes without visible change before the spin. Check the manufacturer's compatibility chart for the plastic and the solvent, and prefer a rated material for solvent work.

**When should a rotor be retired?**

On the manufacturer's schedule based on logged usage, not on appearance. Fatigue is not visible, which is why the log rather than the inspection is the control.

**Can I use any bottle in a large rotor?**

No. Vessels are rated parts with a maximum speed, a compatibility list and a service life, and they are matched to the rotor by part number. Running an unrated or incompatible vessel is the usual cause of a rotor failure.

## Culture formats

Growing cells flat on stiff plastic is convenient and is not neutral. Stiffness, geometry, oxygen tension and the absence of matrix all shape behavior, and results that depend on those properties do not transfer between formats. Choosing a format is choosing which of those distortions to accept.

1. **Name what the format distorts**: Flat culture on stiff plastic changes stiffness sensing, polarity, oxygen exposure and matrix contact. If any of those is central to the question, the format is part of the result and should be stated.
2. **Match gas exchange to the vessel**: Fill volume, baffles, shaking speed and vessel geometry decide oxygen transfer. A flask filled too full is oxygen limited long before it is nutrient limited.
3. **Use a rocking or stirred format for scale**: Beyond a certain volume, static culture cannot supply oxygen. Rocking and stirred formats add mixing and shear, which some cells tolerate and some do not.
4. **Coat and feed primary cells to their protocol**: Primary neural and glial cultures need specific coatings, feeding schedules and often conditioned components. A protocol borrowed from a line will fail quietly.
5. **Measure the culture rather than inspecting it**: Spent medium analysis and automated viability counting turn subjective assessment into numbers that can be compared across weeks and between people.
6. **Consider microfluidic formats for gradients and small numbers**: Where a controlled gradient, a co-culture geometry or very small cell numbers is the point, a chip format answers what a plate cannot, at the cost of throughput and handling.

**Is flat culture a poor model?**

It is a specific model with known distortions: stiffness, geometry and oxygen exposure unlike tissue. That is acceptable when the question does not depend on them and should be stated when it might.

**Why are my cells oxygen limited in a flask?**

Fill volume and shaking, usually. Oxygen enters through the surface, so a deeper culture at the same shaking speed transfers less per cell. Baffles and lower fill volumes are the remedy.

**Do primary neural cultures need special handling?**

Yes: defined coatings, a specific feeding schedule and often supplements that a cell line does not require. Borrowing a line protocol for them is a common cause of failure.

**Should I move my assay to three dimensions?**

Only if gradients, matrix contact or architecture are part of what you are measuring. Otherwise the extra variability and the harder analysis cost more than the added realism is worth.

## Flow cytometry applications

A cell cycle histogram is a measurement of DNA content, and everything that makes it interpretable happens before the acquisition: a stain that binds proportionally, cells in a genuine single cell suspension, and doublets excluded. The fitting model is the last and least important step.

1. **Use a stoichiometric stain and let it equilibrate**: The dye has to bind in proportion to DNA and reach equilibrium. Under staining compresses the distribution and makes the phases overlap, which no fitting model recovers.
2. **Exclude doublets explicitly**: Two cells in one phase pass as one cell with twice the DNA, which is exactly the population being measured. Area against width gating is the standard exclusion and it is not optional.
3. **Ensure a true single cell suspension**: Clumps produce apparent higher ploidy and distort every proportion. Filtering immediately before acquisition, and a gentle dissociation, are what deliver it.
4. **Acquire at a low rate**: High acquisition rates increase coincidence and widen the peaks. For DNA content work, slow acquisition is a requirement rather than a preference.
5. **Watch tandem conjugate degradation**: Tandem dyes lose their acceptor with light and time, which shifts signal into the donor channel and looks like a real population. Compensation controls have to use the same lot and age.
6. **Record instrument settings and the gating tree**: Voltages, the gating strategy and the fitting model together determine the reported proportions. A distribution without them cannot be compared.

**Why are my phases poorly separated?**

Usually the stain has not equilibrated, or the suspension contains clumps, or acquisition was too fast. All three widen the peaks, and none is fixed by a different fitting model.

**Why exclude doublets?**

Because a doublet of cells in one phase has the DNA content of a single cell in the next, so doublets inflate exactly the population you are measuring. Area against width gating removes them.

**Why did my tandem dyes flow cytometry signal move?**

Degradation of the tandem, which releases signal into the donor channel. It is accelerated by light and by time, so controls must match the lot and the age of the reagent used on samples.

**Is flow cytometry analysis software free of charge good enough for a publication figure?**

Yes, provided the gating is described and the raw files are kept. What a reviewer asks about is the strategy and the controls, not the brand of software, and the fitting algorithms in the open tools are the same published ones. Where a regulated report is the output rather than a figure, the question becomes licensing and audit trail rather than capability.

## CMC and clinical development

The manufacturing and control package grows with the program. Early filings need enough to show the material is safe and consistent; later ones need the full characterisation, validated methods and a stability program that supports the claim. Filing too much too early wastes effort; too little delays the trial.

1. **Scale the package to the phase**: Early phase needs identity, purity, potency, safety related impurities and enough stability to cover the study. Later phases add validated methods, process characterisation and a full stability program.
2. **Start real time stability early**: A shelf life claim rests on real time data at the labeled condition. Accelerated data supports it and cannot replace it, so the clock has to start as soon as representative material exists.
3. **Validate the bioanalytical methods to the use**: Methods supporting clinical samples need documented accuracy, precision, selectivity, stability and a defined range before the samples arrive, not afterwards.
4. **Align supply with enrollment**: Batch size, shelf life and release timing have to match the enrollment plan. A slow site and a short shelf life together waste material; a fast site outruns supply.
5. **Keep comparability in mind at every change**: Any change of site, scale or process raises the question of whether the product is the same. Making changes early, while there is little clinical material, is far cheaper than later.
6. **Document the reasoning, not just the results**: Why a specification was set where it was, and what data supports it, is what a reviewer asks about. Recording it as decisions are made is much easier than reconstructing it.

**How complete does an early package need to be?**

Phase appropriate: enough to establish identity, purity, potency and safety for the study duration. Regulators expect it to grow, and over-filing early commits you to specifications you may want to change.

**Can accelerated data support a shelf life?**

It supports and does not replace real time data at the labeled condition. Shelf life claims are extended as real time data accumulates, which is why the program starts as early as possible.

**When should bioanalytical methods be validated?**

Before clinical samples are analyzed. Retrospective validation of a method already used is far weaker and occasionally forces reanalysis of stored samples that may no longer be stable.

## Materials and industrial testing

Engineered materials are tested to standards that specify the specimen as tightly as the measurement, because a property measured on a differently shaped coupon is a different number. Non destructive methods answer a related question without consuming the part, and each has a defined detectable defect population.

1. **Name the standard and the specimen**: Composite properties depend on lay up, fibre direction and coupon geometry, all of which the standard fixes. A result quoted without the standard cannot be compared with anything.
2. **Treat energy storage testing as a hazard program**: Abuse testing of cells involves thermal runaway, gas and fire. Chambers, extraction, suppression and procedures are what make it possible, and few laboratories can host it.
3. **Choose a non destructive method by the defect you expect**: Each method detects a population of defects at a size and orientation. Choosing one without knowing what you are looking for produces a clean report and an undetected flaw.
4. **Inspect used inspection equipment carefully**: Optics, stage condition, illumination and whether the camera and software are supported. The mechanical parts usually survive; the electronics and software usually decide the value.
5. **Apply the same accreditation logic across sectors**: Food, beverage, environmental and materials laboratories are all accredited method by method. The scope schedule is the document to read, whatever the sector.
6. **Specify the report before the work**: Standard, specimen description, conditioning, equipment, result with uncertainty, deviations and a signature. Agreeing it in advance avoids a report that cannot support the decision.

**Why does specimen geometry matter so much?**

Because composite properties are anisotropic and depend on the lay up and the loading direction. Standards fix the geometry precisely so that numbers from different laboratories describe the same thing.

**Can non destructive testing replace destructive testing?**

It answers a different question: whether a defect of a given type and size is present in this part. Destructive testing gives properties. Programs generally need both.

**Is accreditation different in industrial sectors?**

The logic is identical: accreditation is granted for named methods on named materials, published in a scope schedule. Reading it is the same step whether the sample is a composite or a beverage.

## Custom vector production

A custom vector order is a small manufacturing project. The packaging system decides the safety profile and what can be carried, the titre method decides whether two batches can be compared, and the purity decides whether primary cells survive the preparation.

1. **State the packaging system and generation**: Later generation systems split the packaging functions across more plasmids and reduce the risk of a replication competent event. The generation is a safety statement and belongs in the order and in the biosafety application.
2. **Choose transient or integrating deliberately**: An integrating vector gives lasting expression and inserts into the genome. A non integrating adenoviral vector gives high, transient expression without insertion. The experiment's duration decides.
3. **Name the titre method and the cell line**: Functional titre on a cell resembling yours is what predicts transduction. Physical particle counts are much higher and do not. Fix it in the order.
4. **Specify purity for the target cells**: Crude supernatant is adequate for robust lines and toxic to primary cells. Concentration alone concentrates the medium components too, so purification rather than concentration is what primary work needs.
5. **Order the control in the same production**: An empty or non targeting vector from the same run shares the preparation and its contaminants, which is what a control has to do.
6. **Have the approvals before the delivery**: Institutional biosafety approval, containment level and waste route, agreed and documented. This is administrative and it is the commonest cause of a delayed start.

**Concentrate or purify with a virus concentration kit?**

Concentration raises titre and also concentrates medium components, serum and cellular debris. For primary and sensitive cells, purification is what removes the toxicity, and concentration alone frequently makes it worse.

**Integrating or transient vector?**

Integrating for lasting expression through division; adenoviral or other non integrating routes for high, short lived expression without insertion. Duration of the experiment usually decides.

**What has to be in place before delivery?**

Institutional biosafety approval covering the vector and the insert, the containment level, a waste route and trained handlers. All of it is administrative and all of it takes longer than production.

## Vibrational and inspection microscopy

A Raman microscope is a spectrometer that happens to look through an objective, and its images are built one point at a time. That makes acquisition time, not optics, the practical limit, and it makes sample fluorescence the enemy, because a weakly scattering signal sits on top of anything that glows.

1. **Choose the excitation wavelength against fluorescence**: Longer wavelength excitation reduces fluorescence and reduces the Raman signal too. The right compromise depends on the sample, and it is the first decision rather than a setting.
2. **Budget acquisition time realistically**: A map is built pixel by pixel, so an image is the pixel count multiplied by the dwell time. Mapping a large area at high resolution takes hours, and that shapes what experiments are feasible.
3. **Calibrate the wavenumber axis regularly**: Peak positions are the identification, and the axis drifts. A standard reference material run on a schedule is what keeps identifications valid.
4. **Separate inspection questions from chemical ones**: Dimensional measurement, defect detection and surface finish are inspection tasks with their own optics and calibration. A chemical imaging instrument answers them poorly and expensively.
5. **Understand what light scattering sizing reports**: Dynamic light scattering reports a hydrodynamic size distribution weighted by intensity, which emphasises larger particles heavily. It is fast and it is not a count based distribution.
6. **Record the acquisition parameters**: Laser wavelength and power, objective, exposure, accumulations and any baseline correction. A spectrum without them cannot be reproduced or compared.

**Why is my Raman spectrum swamped?**

Sample fluorescence, in most cases. Moving to longer excitation, photobleaching first, or using a time gated approach are the standard remedies, and each costs signal or time.

**How long does a Raman map take?**

Pixel count multiplied by dwell time, which for a decent map at reasonable signal is frequently hours. Planning the area and resolution against available time is part of the experiment design.

**Is light scattering sizing comparable to microscopy sizing?**

No. Scattering reports an intensity weighted hydrodynamic distribution that emphasises large particles; imaging counts particles. They disagree for good reasons and both should state the basis.

## Recombinant protein reagents

Two lots of the same recombinant protein at the same stated mass frequently differ in activity, because folding, aggregation, glycosylation and carrier content differ. Buying on mass and hoping is the default; specifying activity and characterisation is what makes results comparable across a study.

1. **Choose the host for the modification you need**: Bacterial expression gives unmodified protein cheaply; mammalian and insect systems give glycosylation that may be essential for activity or binding. The host is part of the reagent identity.
2. **Demand specific activity, not just purity**: Purity says what fraction is your protein; specific activity says how much of it works. A lot can be pure and largely misfolded, and only the activity figure reveals it.
3. **Check endotoxin and aggregate content**: Endotoxin provokes responses easily mistaken for the protein's effect, and aggregates change potency and can be immunogenic. Both belong on the certificate.
4. **Know whether carrier protein is present**: Carrier stabilises dilute preparations and interferes with some assays and with conjugation. A carrier free version exists for those cases and is a different product.
5. **Titrate every new lot**: Match lots by activity in your own assay rather than by stated mass. It is one dose response and it prevents a step change in the middle of a study.
6. **Measure heterogeneity where it matters**: For proteins whose charge or glycan distribution affects function, a charge variant separation describes the lot in a way a purity figure cannot.

**Why does a new lot behave differently?**

Specific activity differs between lots because folding, aggregation and modification differ. Matching by stated mass rather than by activity is the usual cause of a step change mid study.

**Does the expression host matter for a recombinant human protein if the sequence is the same?**

Frequently yes. Glycosylation and disulphide formation differ between hosts and can change activity, stability and binding, so the host is part of the reagent specification.

**Carrier free or with carrier?**

Carrier stabilises dilute stocks and interferes with conjugation and with some assays. Choose deliberately and keep it consistent, because switching changes effective concentration.

## Alternative chromatography

Reversed phase liquid chromatography solves most problems and is a poor answer to several. A compressible mobile phase changes the viscosity and diffusion that limit speed; a monolithic bed changes the flow path; and phases designed for polar or very hydrophobic analytes address retention that no gradient adjustment fixes.

1. **Use a compressible mobile phase for speed and chirality**: Low viscosity permits high flow at modest pressure, and the solvating power is tuned by pressure as well as by composition. It is the established route for chiral separations at preparative scale and uses far less organic solvent.
2. **Consider monolithic beds for dirty samples and high flow**: A continuous porous bed has a different flow path from packed particles, tolerating high flow at low back pressure. It suits fast separations and samples that would foul a packed column.
3. **Look at multi column operation for repeated separations**: Running columns in sequence with countercurrent movement raises productivity and reduces solvent per gram, at the cost of a more complex system and method development.
4. **Match the phase to polarity extremes**: Very polar analytes are poorly retained on standard reversed phase, and very hydrophobic ones are retained too strongly. Phases exist for both, and switching phase beats adjusting a gradient.
5. **Keep detection compatible with the mobile phase**: A compressible mobile phase and unusual solvents constrain which detectors work and how they are plumbed. Confirm detector compatibility as part of the method choice.
6. **Record the method fully**: Mobile phase composition, pressure, temperature, column and detector settings. These methods are more sensitive to conditions than conventional ones, so the record matters more.

**When is a compressible mobile phase worth it?**

For chiral separations, for preparative work where solvent cost and evaporation dominate, and where speed at low back pressure matters. For routine reversed phase assays it adds complexity for no gain.

**What does a monolithic column change?**

Flow path and back pressure, allowing high flow rates and tolerating particulates better than a packed bed. Selectivity is a property of the chemistry rather than of the format.

**Why will my polar analyte not retain?**

Because it has little to interact with on a hydrophobic phase. A phase designed for polar retention, or a different mechanism entirely, is the answer rather than a weaker gradient.

**What is this technique actually good for?**

Chiral separations, moderately polar compounds that reversed phase handles poorly, and preparative work where solvent consumption matters. It needs a carbon dioxide supply and back pressure control, so it is added for a reason rather than for general use.

## Expression systems

Expression systems are chosen by what the protein cannot do without: disulphide bonds, glycosylation, membrane insertion or simply size. Bacteria are fastest and least capable; insect cells handle complexity at moderate cost; mammalian cells give human like processing slowly and expensively.

1. **List the modifications the protein requires**: Disulphide bonding, glycosylation, phosphorylation and membrane insertion each rule systems in or out. That list, written first, selects the host without further debate.
2. **Separate cloning hosts from expression hosts**: A high efficiency cloning strain is optimized for uptake and plasmid stability, not for expressing protein without degrading it. Using one to express is a common quiet failure.
3. **Budget the insect system's timeline honestly**: Generating and amplifying virus adds weeks before any protein appears, and titre has to be tracked. The capability is real and it is not fast.
4. **Choose the mammalian route for human like product**: Transient expression gives milligrams in weeks; stable lines give more, later. Both cost considerably more per milligram than the alternatives and are the only route for some proteins.
5. **Use a reporter to validate the construct**: A reporter in the same vector confirms transcription, translation and delivery independently of whether your protein folds. It separates construct problems from protein problems.
6. **Characterise before declaring success**: Yield, purity, aggregation state and activity. A system that gives high yield of misfolded protein has not worked, and only the activity measurement reveals it.

**When should I move beyond bacterial expression?**

When the protein needs glycosylation, complex disulphide bonding, membrane insertion or is large and multi domain. Those are folding requirements that induction conditions cannot fix.

**Why not use a cloning strain to express?**

Because it lacks protease deficiency and tight repression, so the protein is degraded or the construct is lost. Expression strains exist for exactly those properties.

**How long does an insect cell campaign take?**

Weeks before protein appears, because virus has to be generated, amplified and titred first. It is a capable system and it should not be chosen when speed is the constraint.

**Why do my insect-cell yields fall over time?**

Usually virus passage. Repeated amplification produces defective particles, so work from a low-passage bank, record the passage number and re-titre rather than carrying a stock indefinitely.

## Expression vectors

A vector backbone carries decisions that outlast the insert: how strongly the gene is transcribed, how cells carrying it are selected, whether it persists and what tags it adds. Those decisions are made once and inherited by every construct built on it, which is why the backbone deserves more attention than it usually gets.

1. **Match promoter strength to the experiment**: A very strong constitutive promoter suits transient overexpression and can be silenced or toxic in a stable line. Weaker or endogenous-like promoters suit long term expression.
2. **Choose the selection marker for the host**: Resistance markers differ in how quickly they kill, in background survival and in whether the host already carries resistance. Kill curves on your cells are what set the concentration.
3. **Decide integration deliberately**: Episomal vectors are lost with division; integrating vectors persist and insert somewhere. Transposon and site specific systems give integration with more control than random insertion.
4. **Add an inducible system only where timing matters**: Induction control costs an extra construct and adds leakiness to manage. It earns that where the product is toxic or where onset timing is the question.
5. **Verify the backbone once, properly**: Whole plasmid sequencing of the empty backbone, kept as a reference. Every construct built on it then inherits verified sequence and only the insert has to be checked.
6. **Record the map and the license terms**: Backbones frequently carry commercial terms that follow every construct made from them. Establish them before a program standardises on one.

**Why is my strong promoter silenced?**

Very strong viral promoters are frequently silenced over time in stable lines, particularly in primary and stem cells. A weaker or more endogenous promoter usually gives lower but durable expression.

**Episomal or integrating?**

Episomal for short transient work with no insertion. Integrating for durable expression, accepting an insertion site and its effects. Transposon or site specific systems sit between them with more control.

**Do I need to sequence every construct?**

Sequence the backbone once and keep it as a reference; then whole plasmid sequencing of each new construct is cheap and catches rearrangements a junction read misses.

## Biosafety cabinets

Biosafety cabinets protect the operator, the environment and the product, and the class determines how much of the air is recirculated and whether volatile chemicals may be used inside. Most bad purchases in this area come from choosing on price and discovering the chemistry restriction afterwards.

1. **Decide the class from the agent and the chemistry**: Recirculating cabinets are unsuitable for volatile or toxic chemicals unless hard ducted, because filters do not remove vapours. Where solvents are used with biological material, the class and the ducting are the decision.
2. **Do not confuse a clean air bench with a cabinet**: A laminar flow bench protects the product and blows air at the operator. It is not a biosafety cabinet, and the names in catalogs make this confusion easy.
3. **Plan ducting and room balance with the building**: A ducted cabinet changes the room's air balance and needs make up air. Installing one without that produces a cabinet that fails certification in place.
4. **Keep the work surface clear and the grilles unobstructed**: Containment depends on airflow patterns that a cluttered surface disrupts. This is the commonest cause of a containment failure in a cabinet that is otherwise fine.
5. **Certify annually and after every move**: Airflow, filter integrity and containment are tested in place, and a cabinet moved to another bench is a different installation. Keep the certificates with the asset record.
6. **Verify the sterilisation route as well**: Waste from the cabinet is autoclaved, and the autoclave has its own verification including leak testing on vacuum machines. The chain is only as good as its weakest link.

**Can I use solvents in a recirculating cabinet?**

Not in any quantity, because the filter removes particles and not vapours, which are returned to the room. Volatile chemistry needs a cabinet class that exhausts, hard ducted to the outside.

**Is a laminar flow bench a biosafety cabinet?**

No. It protects the product by blowing filtered air over the work and offers no operator protection. Using one for infectious material directs the hazard at the user.

**Does a cabinet need recertifying after moving?**

Yes. Certification tests the cabinet in its installed location with that room's airflow, so a move invalidates it. Certify after any relocation as well as annually.

**Why is the same cabinet described in so many ways?**

Because catalogs mix roman and arabic numerals and move the type letter around. Normalise every quotation to class, type letter, ducted or recirculating, and nominal width before comparing; the products are usually identical and the inclusions are not.

**Can I use solvents in a recirculating cabinet?**

Only in the small quantities its type permits, if any. Filters remove particles and not vapour, so a cabinet that returns most of its air to the room is the wrong choice where volatile chemistry is involved. State the chemistry before choosing the type.

## Conjugation chemistry

Conjugation chemistry is judged on selectivity: whether the reaction goes where you intended and nowhere else. Random coupling to abundant residues gives a heterogeneous product; a designed attachment point gives one species that can be characterised and reproduced.

1. **Design the attachment point before the chemistry**: An enzymatically installed tag, an unnatural amino acid or an engineered residue gives a single defined site. Random coupling to abundant residues gives a distribution that has to be characterised every batch.
2. **Choose copper free chemistry for living systems**: Copper catalysis is efficient and toxic to cells. Strain promoted variants proceed without it, more slowly, and are the route for live cell and in vivo work.
3. **Start from characterised material**: A conjugation on a peptide of unknown purity or net content produces an uninterpretable mixture. Establish purity, net peptide content and identity first.
4. **Plan the purification of the conjugate**: The product has to be separated from unreacted starting materials and from over-coupled species, and the yield is frequently modest. Budget the separation and the analytics.
5. **Characterise the conjugate, not the components**: Mass, degree of labeling and its distribution, aggregation state and retained activity. A conjugate described only by its intended structure has not been characterised.
6. **Order named ligands as reagents**: A well known signalling ligand ordered as a characterised protein comes with activity data. Ordering the sequence and making it yourself adds a folding project to the experiment.

**Why use copper free chemistry?**

Because copper catalysis is toxic to cells and can damage proteins. Strain promoted reactions are slower and proceed without it, which is what makes live cell and in vivo labeling possible.

**Random or site specific coupling?**

Site specific wherever the conjugate has to be reproducible or characterised, because it gives one species. Random coupling is faster, cheaper and gives a distribution that must be measured each time.

**What should a conjugate certificate state?**

Mass confirmation, degree of labeling and its distribution, purity after conjugation, aggregate content and retained activity. Purity of the starting material alone is not enough.

## Genome sequencing applications

Depth is the main cost lever in genome sequencing and the main determinant of what can be concluded. Shallow coverage across the whole genome answers questions about large scale structure cheaply; calling individual variants, and especially rare ones in a mixture, needs depth and molecular tagging.

1. **Match depth to the smallest thing you must detect**: Copy number and ancestry survive shallow coverage. Germline variant calling needs moderate depth. Rare somatic variants in a mixture need deep coverage plus molecular identifiers to separate signal from error.
2. **Judge library chemistry on evenness**: Coverage uniformity and duplicate rate matter more than nominal yield, because uneven coverage leaves regions uncalled at any average depth.
3. **Use molecular identifiers where the fraction is small**: Tagging original molecules distinguishes a true low frequency variant from an amplification or sequencing error. Without them, low frequency calling is guesswork.
4. **Keep a targeted method for targeted questions**: When the variant is known and the question is present or absent, amplification is faster and cheaper than sequencing. Not everything needs a genome.
5. **Plan storage and analysis before generating data**: Genome scale data is large and long lived. Where it lives, who pays after the project and how it is analyzed reproducibly are decided at the start or not at all.
6. **Agree deliverables including the pipeline**: Raw reads, alignments, variant calls with their filters, the reference and annotation versions, and runnable code. A variant table alone cannot be re-analyzed.

**What can shallow coverage answer?**

Copy number, large structural features and ancestry, at a small fraction of the cost. It cannot reliably call individual variants, and describing it as whole genome sequencing without the depth caveat overstates it.

**Why do rare variants need molecular identifiers?**

Because at low allele fraction the true signal is comparable to the error rate. Tagging original molecules lets errors introduced after tagging be identified and removed.

**Is sequencing always better than amplification?**

No. For a known variant with a yes or no answer, a targeted amplification assay is faster, cheaper and easier to validate. Sequencing earns its place when the question is open.

## Stereo and dissecting microscopes

A dissecting scope is the instrument you buy when the specimen has to stay under your hands. It trades magnification for a long working distance, an upright image and enough depth of field that a moving, three-dimensional object stays in focus while you cut, sort or position it. Almost every buying mistake here comes from comparing these instruments on magnification, which is the one specification that matters least for the work they are bought to do.

1. **Working distance, before anything else**: Working distance is the gap between the front lens and the specimen, and it decides whether your hands, forceps, a pipette or a soldering iron fit under the objective at all. A long working distance head costs more and magnifies less. Buy the distance the task needs first, then see what magnification is left, because no amount of optical quality rescues an instrument you cannot get a tool underneath.
2. **Zoom range against fixed steps**: A zoom body gives a continuous range and a click-stop body gives two or three fixed powers. Continuous zoom is worth paying for when the operator reframes constantly during a dissection or an inspection; fixed steps are more repeatable and cheaper to keep parfocal, which matters when several people share the instrument and each expects the same view at the same setting.
3. **Head configuration and who else has to see the image**: A binocular head serves one operator. A trinocular head adds a vertical port for a camera and is the only configuration that lets you document without losing the eyepieces. Decide this at purchase: retrofitting a camera port to a binocular body is usually impossible, and adapters that split the eyepiece light cost you the view you were trying to record.
4. **Illumination, which is half the instrument**: Transmitted light shows outlines through translucent samples, oblique incident light shows surface texture, and a ring light flattens shadows for photography. Most stereo work needs at least two of these and a dimmer on each. A stand quoted without its illuminator is not a cheaper stand, it is an incomplete one, and the lighting is often the larger part of the difference between two quotes.
5. **Stand and base for the specimen you actually have**: A pillar stand suits small dishes, a boom stand suits large or awkward objects that cannot be brought to the instrument, and a gliding arm suits work that moves across a tray. The stand is the part that gets outgrown first, so buy the reach for the biggest object the bench will ever see rather than the one on the bench today.

**What is the difference between a dissecting light microscope and a compound one?**

Optical path and purpose. A dissecting instrument uses two separate light paths to give an upright, three-dimensional view of a whole object at low power with a long working distance. A compound instrument uses one path at high power through a thin prepared sample. You choose by whether you need to work on the specimen or look inside it.

**Do I need zoom or are fixed powers enough?**

Fixed powers are cheaper, more repeatable and perfectly adequate for a single repeated task. Zoom earns its price when operators reframe constantly or when one instrument has to serve several kinds of work, and it is worth confirming the zoom stays in focus across its whole range before you buy.

**Can I add a camera to a binocular head later?**

Usually not well. Eyepiece adapters lose field and shift focus, so if documentation is likely at any point, specify a trinocular body now. The price difference at purchase is far smaller than the cost of replacing the body when the imaging requirement arrives.

**Does the illuminator come with the stand?**

Not always, and this is the most common reason two quotes look different. Ask for the light source, its type and its dimmer to be listed as line items, and confirm whether transmitted, incident or both are included before comparing prices.

**Is a stereo microscope with camera cheaper than adding one later?**

Usually, and it is also less likely to disappoint, because the bundled camera is matched to the port and the illumination. Adding one later is straightforward on a trinocular head and awkward on a binocular one, where an eyepiece adapter costs magnification and light. If a camera is likely within a year, buy the trinocular head now.

**How should a stereo zoom microscope specification be read?**

Three numbers decide it: the zoom range, the working distance at the highest magnification you will use, and the field of view at the lowest. Everything else is convenience. A specification that quotes total magnification without the eyepieces it assumes is not comparable to another one.

## Inspection microscopes

Inspection is the one microscopy application where the answer has to survive being disagreed with. A finding that stops a shipment, fails a batch or supports a warranty claim is read by someone who was not at the eyepiece, so the instrument has to produce a record as well as a view. That single requirement changes what is worth buying, and it is the reason inspection benches drift from optical heads to digital ones as soon as the findings start leaving the room.

1. **Decide whether the output is a decision or a document**: If an operator looks, decides and moves on, an optical stereo head is faster and cheaper and nothing beats it for hand-eye work. If the finding has to be attached to a report, sent to a supplier or compared against last month's, you are buying an imaging system that happens to have optics, and the camera, the software and the storage matter as much as the lens.
2. **Working distance against the fixture**: Inspection samples arrive in jigs, trays and carriers that are taller than anyone expects. Measure the tallest fixture the bench will see and specify working distance against that, not against the part. This is the specification most often got wrong, because the part is what people bring to the demo.
3. **Repeatability between operators**: Two inspectors reaching different conclusions is an instrument problem as often as a training one. Click-stop magnification, a fixed illumination geometry and saved camera settings make the same defect look the same to everyone. Continuous zoom and a hand-held light are flexible and much harder to make repeatable across a shift.
4. **Lighting geometry for the defect class**: Scratches and tool marks show under oblique light, contamination shows under diffuse light, and cracks in transparent parts show under transmitted light. Decide the defect classes first and buy the geometries that reveal them. A ring light alone flattens exactly the surface texture that most mechanical inspection is looking for.
5. **Measurement, if a number has to leave the bench**: If findings are recorded as sizes rather than descriptions, calibration becomes part of the purchase: a stage micrometer, a documented calibration routine and software that stores the scale with the image. Without that, a measured number on an inspection report is an opinion with a decimal point on it.

**Optical or digital inspection microscope for incoming inspection?**

Optical if the operator decides at the bench and the finding stays there. Digital if the finding is reported, shared or compared over time, because the record is then the product of the bench and an optical head produces none.

**Does a ring light suit all inspection work?**

No. Ring lighting is even and flattering, which is exactly wrong for surface defects. Scratches, dents and tool marks need oblique light; buy the geometry that shows your defect class rather than the one that makes parts look best.

**Can one instrument cover inspection and particle counting?**

Rarely. Particle work runs on filter membranes with an automated stage and classification software and is specified as a method, not as a stand. If both are needed, price them as two systems from the start.

## Measuring and toolmakers microscopes

Any microscope can display a number. Very few produce one that survives being questioned. The difference is not the optics, which are usually unremarkable on a metrology instrument, but the stage, the scale, the calibration routine and the record that ties a reading to a traceable standard. Buying on image quality is the standard mistake in this category, and it produces an instrument that looks better and measures worse than the one it replaced.

1. **The stage, not the objective**: On a metrology instrument the measurement is made by moving the specimen a known distance, so stage accuracy, repeatability and squareness set the error budget. A crisp image on a loose stage produces a confident wrong answer, which is worse than a soft image on a good one because nobody questions it.
2. **How the scale is read**: An eyepiece graticule read by eye, a mechanical micrometer drum and a linear encoder feeding a digital readout are three different levels of resolution, repeatability and operator dependence. Each is adequate for some work and useless for other work, and the choice should follow the tolerance you have to prove rather than the budget you happen to have.
3. **Edge definition and lighting**: Every measurement of a part is a measurement of where its edge appears to be, and that depends on illumination. Profile measurement under transmitted light gives a hard silhouette; surface measurement under incident light gives an edge that moves with the lighting angle. Fix the geometry before you argue about the reading.
4. **Calibration and traceability**: A measuring instrument needs a documented calibration against a certified standard, at a stated interval, with the certificate kept. If the readings support a customer specification, an audit will ask for that chain. Budget for the artefact and the recurring calibration at the same time as the instrument, because retrofitting traceability is the expensive way to do it.
5. **Operator dependence**: A reading that changes with who takes it is a process problem the instrument can reduce. Crosshair alignment by eye varies between people; edge detection in software does not. If several inspectors produce the same report, that argument for automated edge detection is usually stronger than the argument for better optics.

**Is an eyepiece graticule a measurement?**

It is, once it has been calibrated against a stage standard at the magnification in use, and provided the tolerance you are judging is well clear of the operator's judgement of where the line sits. Below that it is an estimate presented as a number.

**What matters most on a tool makers microscope?**

Stage accuracy and repeatability, then how the scale is read, then illumination geometry. Optical quality matters least: these instruments exist to move a part a known distance, not to resolve fine detail.

**How often does a measuring instrument need calibrating?**

At an interval you can defend, documented, against a certified artefact, and immediately after any move or repair. The interval belongs in the quality system rather than in the instrument specification, and the recurring cost should be priced with the purchase.

## Clinical and pathology microscopes

A clinical instrument is used for hours at a time by people who cannot afford to be tired or to see a color slightly wrong. That changes the specification away from headline resolution and toward the things that decide whether a long session stays accurate: the height and angle of the head, the evenness and color temperature of the illumination, and how quickly the operator can move between powers without refocusing. These are the specifications that get skipped in a demo and regretted for a decade.

1. **Ergonomics for a full session**: An adjustable-height, tilting head and a correctly set eyepiece position are the difference between a comfortable day and a musculoskeletal complaint. Everyone who will use the instrument should sit at it before purchase. The people who use it hardest are usually not the people who attend the demo, which is how the wrong head configuration gets signed off.
2. **Illumination and color rendering**: Stained material is read by color, so the light source has to render color consistently across the field and across its dimming range. Uneven illumination and a source whose color shifts as it is dimmed both change the appearance of a stain, which is a quality problem rather than a comfort one.
3. **The objective set that matches the work**: Routine screening, high-power review and oil-immersion work each need their own objective, and a parfocal, parcentric set that holds focus and center through a rotation saves minutes on every case. Specify the set by the tasks on the bench rather than accepting the vendor's standard turret, which is assembled to a price.
4. **Documentation and second opinions**: If images leave the room, a trinocular port and a camera with faithful color reproduction are part of the instrument, not an accessory. Color accuracy matters more than pixel count here, because a second reader is judging a stain rather than counting structures.
5. **Service and the life of the instrument**: Clinical stands are kept for a long time, so the questions that matter are parts availability, service response and whether the illumination is a standard replaceable module. Ask what the instrument costs to keep running for a decade; the purchase price is usually the smaller number.

**What makes an instrument suitable for pathology sign-out?**

Ergonomics, even illumination with consistent color rendering, a parfocal objective set covering scanning through high power, and a comfortable stage. Resolution beyond that is rarely the constraint; operator fatigue is.

**Is a dual head teaching microscope worth it for training?**

For one-to-one teaching and consensus review, yes, because both people see the same field with a shared pointer. For showing a room, a camera and a screen is cheaper, records the session and reaches more people.

**Does hematology need a different instrument from tissue work?**

It needs a different emphasis. Film review is high-power oil work rewarding a good immersion objective and smooth one-handed stage control; tissue review spends its time scanning at lower power and rewards field of view and fast power changes.

## Polarising microscopes

Polarised light turns an optical instrument into a measuring one. The colors and extinctions it produces are used to identify a material, confirm a crystal form or find a contaminant, and that only works if the instrument itself introduces nothing. Ordinary optics are not strain-free, ordinary stages do not rotate about the optical axis, and an instrument assembled from ordinary parts will produce convincing artefacts that look exactly like findings.

1. **Strain-free optics throughout**: Every glass element between the polariser and the analyser has to be free of internal stress, because stressed glass is itself birefringent and adds its own color to the image. This is the single specification that separates a genuine polarising instrument from a standard stand with two filters added, and it is not visible in a photograph of the instrument.
2. **A centred, rotating stage**: Extinction angles and orientation effects are read by rotating the specimen about the optical axis, so the stage must rotate smoothly and stay centred through a full turn. A stage that wanders takes the feature off the crosshair and makes angle measurement impossible rather than merely inconvenient.
3. **A rotatable, calibrated analyser**: Crossed positions have to be exact and repeatable, and the analyser needs to be removable for ordinary viewing. Detents that put it reliably at the crossed position, and a scale for intermediate angles, are the difference between an instrument that measures and one that only demonstrates.
4. **Compensators for the measurement you need**: Retardation is read against a known standard, so the compensator set follows the question: a full-wave plate for sign of elongation, a quarter-wave plate for small retardations, a tilting compensator for quantitative work. Buy the compensators with the instrument; sourcing them later for an older stand is often the hardest part.
5. **Bertrand lens and conoscopy, if you need it**: Interference figures viewed at the back focal plane distinguish uniaxial from biaxial materials and give sign and optic angle. That requires a Bertrand lens and a high-aperture condenser. If mineral or crystal identification is the job, this is not optional; if the work is contaminant screening, it usually is.

**Can I add polarisers to an ordinary microscope?**

For demonstration, yes. For a result you will report, no: ordinary optics carry internal strain that is itself birefringent, and the instrument then adds color of its own. Strain-free optics and a centred rotating stage are what make the technique quantitative.

**What does a compensator actually do?**

It adds a known, calibrated retardation to the specimen's own, so the unknown can be read against a standard. Which plate you need follows the measurement: sign of elongation, small retardations or quantitative values each call for a different one.

**Is differential interference contrast microscopy a form of polarised light microscopy?**

It uses polarised light but answers a different question. Crossed polars and compensators identify materials by their optical behavior; interference contrast makes transparent, unstained specimens visible and its relief is an optical effect rather than real topography.

## Microscope cameras

A camera on a microscope is not a camera on a tripod. The optics in front of it were designed to fill a circular field at a fixed distance, and every decision about sensor size, mount and adapter changes how much of that field reaches the sensor and how faithfully. The commonest disappointment in this category is an image that is sharper in the eyepieces than on the screen, and it is almost always a mounting problem rather than a sensor one.

1. **Sensor size against the field the optics deliver**: The optics project a circular image of a fixed diameter. A sensor larger than that circle records black corners; a sensor much smaller than it crops the field and makes the camera view a fraction of what the eyepieces show. Match the sensor diagonal to the field the instrument delivers, then choose resolution, rather than the other way round.
2. **The mount and the adapter's magnification factor**: The standard thread on a research body is one mechanical interface, and between it and the sensor sits an adapter with its own magnification factor. That factor is what actually sets the recorded field, so an adapter is part of the optical specification, not a fitting. Ordering a body, a camera and no adapter is the most common way to end up with a cropped image.
3. **Pixel size against what you are measuring**: More pixels do not add detail the objective did not resolve. Beyond the point where pixels are small relative to the resolved detail, extra resolution costs sensitivity and speed and buys nothing. For quantitative or low-light work, larger pixels and lower noise beat a bigger number on the specification sheet.
4. **Speed, and what you lose to get it**: Fast acquisition matters for live specimens, moving parts and anything where a shutter interval blurs the subject. Frame rate is usually bought with reduced field, higher noise or reduced bit depth. Decide which of those you can afford to lose before comparing quoted rates, because the rates are quoted at different settings by different suppliers.
5. **Software, licensing and where images live**: Capture software decides exposure control, white balance, scale bars, measurement and file format, and it is where long-term lock-in sits. Ask whether the license is per seat or per camera, whether the image format is open or proprietary, and what happens to stored images if the camera is later replaced by a different make.

**Why does my camera see less than the eyepieces?**

Because the adapter's magnification factor and the sensor size together set the recorded field. A smaller sensor or an adapter with the wrong factor crops the circle the optics project. Match sensor diagonal and adapter factor to the instrument's field before comparing cameras.

**Is more resolution always better?**

No. Once pixels are small relative to what the objective resolves, more of them add noise and slow acquisition without adding detail. For dim or moving specimens, larger pixels and lower read noise are the better purchase.

**Colour or monochrome for a CCD microscope camera?**

Color for stained material read by eye, where fidelity is the point. Monochrome for intensity measurement and low light, because a color filter array discards most of the light before it reaches the sensor.

## USB and wireless microscope cameras

How a camera connects decides more of the daily experience than the sensor inside it. Bus-powered cameras are simple until the cable has to reach a second bench; wireless removes the cable and adds latency that makes focusing unpleasant; a direct video output drives a screen with no computer at all but records nothing. None of these is better, and the right answer follows from where the image has to end up rather than from the specification sheet.

1. **Bus power and cable length**: A camera powered from the data cable is one item to carry and one thing to fail. The limit is cable length and the power the port can supply, and extending it with a passive cable is the usual cause of dropped frames that look like a software fault. Where the computer cannot sit beside the instrument, plan for a powered extension from the start.
2. **Latency, and what it does to focusing**: Focusing is a feedback loop between hand and image. Any delay between moving the stage and seeing the result makes fine focus difficult and fine positioning worse. Wireless links and some compression modes add enough delay to be noticeable, which is tolerable for viewing and irritating for acquisition.
3. **Drivers, operating systems and lifespan**: The camera outlives the computer it was bought with. Ask which operating systems are supported, whether the driver is maintained, and whether the camera also presents as a standard imaging device so it keeps working when the bespoke software stops being updated. This is the failure mode that retires otherwise good cameras.
4. **A direct video output for a room**: An output that drives a monitor with no computer in the path is the simplest way to put an image in front of several people. It is fast and reliable and it stores nothing, so if the session has to be recorded, a capture device or a second path to a computer has to be specified alongside it.
5. **Integrated bodies against separate cameras**: An instrument with the camera built in removes the adapter question and guarantees the sensor matches the optics. It also fixes the sensor for the life of the instrument, which is the trade. Integrated suits routine documentation; separate suits laboratories that expect the imaging requirement to change.

**Is a wireless microscope camera good enough for routine work?**

For viewing and sharing, yes. For focusing and acquisition it depends on latency: any perceptible delay between moving the stage and seeing the change makes fine work unpleasant. Try it on your own instrument before committing.

**Do I need a computer at the bench?**

Not if a direct video output to a monitor covers the requirement, which it often does for shared viewing. You do need one, or a capture device, the moment images have to be stored, measured or attached to a record.

**Is an integrated camera body a compromise?**

It is a trade. It removes the adapter and field-of-view problem and fixes the sensor for the life of the instrument. That suits routine documentation and suits a laboratory expecting its imaging needs to change much less well.

**Is a digital microscope for sale second hand worth considering?**

The optics and the stand are worth buying used and the camera usually is not, because the sensor is the part that dates and the driver support is the part that disappears. Check that the model's software still runs on a current operating system before anything else; an instrument whose camera cannot be driven is a stand.

**What does a digital video microscope add over stills?**

Frame rate, and with it the ability to record something moving: a beating culture, a dissection, an injection. It costs resolution per frame and a great deal of storage, so it is a mode to use deliberately rather than a better camera. Check the frame rate at the resolution you would actually record at, not at the lowest one.

**Is a wireless digital microscope camera a compromise?**

For routine documentation it is fine and it removes a cable from a crowded bench. For anything timed or measured it is not, because the latency is variable and the compression is usually fixed in the camera. Where an image will be quantified, keep the wire.

## Digital optical microscopes

Replacing eyepieces with a screen changes who can use the instrument and what it produces. Several people can see the same field, posture stops being a constraint, and every observation can be captured. What is given up is resolution at the limit, dynamic range as the eye perceives it, and the immediate, latency-free feedback that makes fine focusing quick. Whether that trade is worth making depends entirely on the work.

1. **The optics are still the instrument**: A screen cannot show detail the objective did not resolve. Numerical aperture, objective quality and illumination set the ceiling exactly as they do on an optical stand, and a large screen fed by a modest objective produces a big, soft picture. Judge the optics first and the display second.
2. **Screen size, resolution and viewing distance**: A display only helps if the detail it shows is visible from where the operator sits. A high pixel count on a small panel at arm's length is wasted, and a large panel with a low count is worse. Check the combination at the working distance people will actually use, not at the demo bench.
3. **Responsiveness while focusing**: Delay between the focus knob and the screen is the single biggest complaint about these instruments. Ask to focus on a real specimen rather than watching a recorded demonstration, because compression and processing that look fine on a static image make fine focus frustrating.
4. **Where the images go**: Storage on the instrument is convenient and a liability: it fills, it is not backed up, and it is rarely accessible from anywhere else. Ask how images leave, in what format, and whether acquisition settings travel with them, because an image with no settings cannot be compared with another.
5. **Depth composition and true three-dimensional work**: Some instruments stack images through focus to produce an everywhere-sharp picture, and some also produce height data from that stack. The first is a photographic convenience; the second is a measurement and needs calibration and a stated uncertainty. They are sold under similar language and they are not the same purchase.

**Is a screen instrument as good optically?**

The optics are the same question as on any stand. What changes is the display path: resolution at the limit and perceived dynamic range are usually a little worse, and shared viewing, posture and capture are much better.

**Can I measure from a high resolution digital microscope's screen image?**

Only with a calibration made at that magnification on that instrument and stored with the image. Measuring from a screen with a variable standoff, as on a hand-held instrument, is not a measurement.

**What should I check on a used digital instrument?**

The software's supported operating systems, whether images export in an open format, what the refurbisher replaced, and whether the display and sensor are original. Those decide how many years are left in it far more than the optics do.

## Microscope illumination

Illumination decides what a microscope can show at least as much as the objective does. Two instruments with identical optics produce different images under different light, and most complaints about contrast, color or missing detail turn out to be lighting problems. It is also the part most often left off a quote, which is why two prices for what looks like the same instrument can be so far apart.

1. **Geometry before brightness**: Light through the specimen shows internal structure, light onto it at a shallow angle shows surface texture, and light from all around shows shape without shadow. Deciding which of these the work needs settles most of the specification, and no amount of brightness substitutes for the wrong geometry.
2. **Color temperature and its stability**: Anything judged by color needs a source whose color does not shift as it is dimmed. Some sources hold color across their range and others warm noticeably, which changes the appearance of a stain or a coating between operators. Ask specifically whether dimming is by current or by a method that holds color.
3. **Dimming, and its repeatability**: A dimmer with no scale is a setting nobody can return to. Where images are compared over time, a numbered or stepped control is worth more than a smooth one, because it makes the illumination part of the record rather than part of the operator's judgement.
4. **Replacement and running cost**: A lamp that is still manufactured, in a standard fitting, held in stock is worth more than a marginally better one that is not. Solid-state sources last far longer but are often a sealed module rather than a replaceable part, which changes the repair from a consumable to a service visit. Ask which it is before buying.
5. **Heat, and what it does to the specimen**: Warm sources heat the stage and the sample, which matters for living material, volatile preparations and long exposures. Solid-state sources run cooler and can usually be switched off between observations without a warm-up penalty, which is a practical advantage on a busy bench.

**Are solid-state sources always better?**

They last longer, run cooler and hold color better when properly driven. They are worse when they are a sealed module that cannot be replaced by the user, and a retrofit into a housing designed for a filament can illuminate the field unevenly.

**Why does my image look flat under microscope lighting?**

Usually because the light is even and frontal. Surface features are visible by the shadows they cast, so shadow-free illumination hides them. Switch to oblique light, or to a segmented ring where only part is lit.

**Is the illuminator included in the price?**

Often not, and it is the commonest reason two quotes differ. Ask for the source, supply, dimmer and filters as line items, and confirm which geometries are covered before comparing.

## Electron microscopy services

Electron and scanning probe instruments are the clearest case in the laboratory where the purchase decision is really a decision about staff and space. The instrument is a fraction of the true cost: preparation equipment, a suitable room, a service contract and, above all, a person who can prepare specimens and interpret images are what decide whether an owned instrument produces answers or photographs.

1. **How many samples, and how predictably**: A steady, high volume of similar samples favors ownership because the marginal cost of each answer falls. Sporadic work on varied samples favors a service laboratory, which already has the preparation routes, the standards and the interpretation experience for material it sees constantly.
2. **Who prepares the specimen**: Preparation is most of the skill and most of the failure. Coating, sectioning, thinning and mounting each have their own equipment and their own learning curve, and an instrument bought without that capability produces poor images from good samples. This is the question that sinks most owned instruments.
3. **What the answer has to be**: Surface shape and composition, internal structure at high resolution, and true surface height are three different answers from three different instrument families. Deciding which answer is needed usually narrows the choice faster than any comparison of specifications does.
4. **Environment and siting**: Vibration, stray magnetic fields, acoustic noise and temperature stability all limit what an instrument achieves, and a site survey before purchase is cheaper than remediation afterwards. A service laboratory has already solved this, which is part of what its fee buys.
5. **The full cost of running it**: Service contracts, consumables, preparation equipment, room conditioning and staff time are the recurring costs, and together they usually exceed the amortised purchase. Price them over the expected life before comparing with a per-sample service fee.

**Which electron instrument do I need?**

If the question is about a surface, its texture or its composition, the scanning family and modest preparation. If it is about internal structure at high resolution, the transmission family and specialized thin-section preparation, which is the part most laboratories send out.

**Is owning cheaper than sending samples out?**

Only at a steady, high volume with in-house preparation and interpretation. Count the service contract, consumables, preparation equipment, room conditioning and staff time over the instrument's life before comparing with a per-sample fee.

**What does a probe instrument add?**

True height data and mechanical or electrical properties, in air or liquid, on specimens that would not survive a vacuum. It answers questions about surface topography and material behavior that an image of brightness cannot.

**Which of the two do I need?**

Surface topography, texture or composition on a bulk sample: the scanning family, with modest preparation. Internal structure at high resolution: the transmission family, with specialized thin-section preparation that is usually sent out.

## Types of microscopy

Almost every disappointing microscope purchase is a class error rather than a model error. The buyer compares two instruments carefully and neither can do the job, because the job needed a different optical principle altogether. Narrowing to the class first takes an afternoon and removes most of the risk, and it is the step that gets skipped when a supplier is asked for a recommendation before the question has been written down.

1. **Is the specimen whole, or thin and prepared?**: A whole object that has to stay intact, be handled or be worked on needs a low-power instrument with a long working distance and an upright, three-dimensional view. A thin, prepared, transparent specimen needs a high-power instrument with a condenser underneath. This single question separates the two most common laboratory classes.
2. **Is the contrast already there, or does it have to be made?**: Stained material carries its own contrast. Living, unstained material carries almost none, and the instrument has to create it optically through phase, interference or related methods. Buying a standard brightfield stand for unstained live work is the classic mismatch, because the specimen is genuinely there and genuinely invisible.
3. **Is the signal light emitted, or light transmitted?**: Fluorescent labels emit their own light, which needs an excitation source, matched filters, a sensitive detector and a dark room rather than a brighter lamp. That is a different instrument, not an accessory, and the filters are chosen from the labels before anything else is specified.
4. **How small is the smallest thing you must see?**: Light-based instruments have a physical resolution limit set by the wavelength. Below it, the question moves to electron or scanning probe instruments, which change the specimen preparation, the operating environment and the skill required. If the answer is near that limit, decide early, because the two routes share almost nothing.

**How many microscope classes does a laboratory really need?**

Usually two: a low-power instrument for whole objects and handling, and a high-power instrument for prepared specimens. Cell work adds an inverted stand and labeled work adds a fluorescence path, and both are separate instruments rather than accessories.

**Can one instrument cover several classes?**

A modular research stand can carry several contrast methods, which is why it costs what it does. It cannot cover the low-power handling job and the high-power prepared-specimen job at once, because those need opposite working distances.

**When do I leave light microscopy altogether?**

When the smallest feature you must resolve approaches the limit set by the wavelength of light. Below that the answer is an electron or scanning probe instrument, with different preparation, different siting and a different skill set.

## Buying a laboratory microscope

Microscope listings are written to be compared on the wrong things. Two instruments with the same body name can differ by several times in price because of objectives, head, illumination and stand, and none of those is usually in the headline. The buyer's job is to convert a listing into a full configuration, because until that is done the prices are not comparable and neither are the instruments.

1. **Get the configuration as line items**: Body, head, eyepieces, each objective by its own specification, condenser, stage, illumination and any camera adapter. Objectives are the largest variable and the easiest to leave vague, and a stand quoted with a basic set beside one quoted with a research set is not the same purchase at all.
2. **Establish what has been serviced, and by whom**: On a refurbished instrument, ask what was cleaned, what was replaced and who did it. Optics cleaning, new illumination and a serviced focus mechanism are the three that matter. An instrument sold as working and an instrument sold as serviced are different products at similar prices.
3. **Check parts and service availability**: The question that decides the instrument's useful life is whether the manufacturer still supports it and whether a local engineer will. An unsupported instrument can still be excellent for a decade and is worth less, and knowing which situation you are in is part of the purchase rather than an afterthought.
4. **Price the whole obligation**: Delivery, installation, alignment, a service visit, consumable illumination and any software license belong in the comparison. On refurbished instruments these are frequently excluded and are the reason a low headline price ends up level with a new one.
5. **Decide whether you are buying or renting the capability**: A short project, an unproven requirement or a peak in workload is often better served by renting, which converts a capital argument into an operating one and returns the instrument when the work ends. Rental also lets a laboratory try a class of instrument before committing, which is the cheapest way to avoid a class error.

**Is a refurbished laboratory microscope, or used microscopes in stock, a good buy?**

Often, because the optics and mechanics age slowly. Establish what was serviced and by whom, whether parts and service are still available, and whether illumination has been replaced. Those three answers decide the value far more than the age does.

**When does renting make more sense?**

For a defined project, an unproven requirement or a temporary peak in workload, and as a way to try a class of instrument before committing capital to it. It converts a capital decision into an operating one and returns the instrument when the work stops.

**Why do quotes for the same instrument differ so much?**

Because the body is a small part of the price. Objectives, head, condenser, stage and illumination make up most of it and are often left vague in a listing. Ask for line items and the prices usually become comparable.

## Biological microscopes

The bench instrument in a biology laboratory does more different jobs than any other microscope in the building, which is why it is so often bought badly. It has to handle stained slides, wet mounts, counting chambers and whatever arrives unannounced, and the specification that serves all of those is decided by the objectives and the condenser rather than by the body or the total magnification printed on the box.

1. **The objective set, in the order you will use it**: A low-power objective for finding the specimen, a medium one for most work, a high dry objective for detail and an oil objective where the specimen demands it. A parfocal, parcentric set saves a refocus on every change, and buying the set rather than accepting the standard turret is the single most useful decision in the purchase.
2. **The condenser, which decides contrast**: Light reaching the specimen at the right angle is what makes detail visible, and an adjustable condenser with an iris is what controls it. Instruments sold with a fixed condenser look cheaper and give a flatter image that no objective can rescue. If unstained or live material is in scope, a condenser with phase annuli belongs in the specification.
3. **Magnification claims against real resolution**: Total magnification is the product of objective and eyepiece and says nothing about resolving power, which is set by numerical aperture. Beyond a point, extra magnification enlarges an image without adding detail, and a listing that leads with a large magnification figure and omits numerical aperture is describing the wrong thing.
4. **Head configuration for who uses it**: Two eyepieces for one observer is the default. A third tube for a camera is worth specifying whenever documentation is plausible, because it cannot be added later to most bodies. A second observation head is a specific purchase for teaching and joint review, and it changes the body, not just the fitting.
5. **Live and cultured material changes the geometry**: Cells growing in a flask or a plate cannot be inverted onto a slide, so they need an instrument that looks up through the vessel. That is a different stand with different objectives corrected for the vessel bottom, and it is the requirement most often discovered after an upright instrument has already been bought.

**Does more total magnification mean more detail?**

No. Resolving power comes from the objective's numerical aperture. Past the point where the eye can see what the objective resolves, extra magnification enlarges the same detail and adds nothing, which is why a listing leading with a large magnification figure is describing the wrong specification.

**Do I need an oil objective?**

Only where the detail you need is beyond what a dry objective can resolve, such as bacterial morphology or fine cell detail. It requires the correct medium, the coverslip thickness the objective was corrected for, and cleaning after every session.

**Can an upright instrument be used for cell culture?**

Not usefully. Cells in a flask or plate have to be viewed from below through the vessel, which needs an inverted stand with objectives corrected for the vessel bottom. This is the requirement most often found too late.

**What is an electric microscope?**

Usually one of three things: a mains-powered bench instrument with built-in illumination, a digital instrument with a screen, or an electron microscope misremembered. Asking what the specimen is resolves which you need.

## Microscope accessories and consumables

An instrument performs to its specification only when everything in the light path matches what its designers assumed. Coverslip thickness, immersion medium, slide flatness and stage stability are all part of that path, and each is bought as a consumable by someone who is not thinking about optics. This is why an expensive objective can produce a disappointing image with nothing visibly wrong.

1. **Coverslip thickness, which the objective assumes**: High-power dry objectives are corrected for one coverslip thickness, and departing from it degrades the image in a way that looks like a focus problem. Buy the graded thickness the objectives specify rather than the cheapest box, and keep one grade on the bench so nobody has to choose.
2. **Slide quality and flatness**: Slides vary in thickness, flatness and cleanliness, and on an automated or high-power stage that variation shows as focus drift across the field. Pre-cleaned, ground-edge slides of a stated thickness cost slightly more and remove a whole class of intermittent problem.
3. **Adhesion and charge for the section that has to stay on**: Sections and cells lift during staining unless the surface holds them. Charged or coated slides exist for exactly that, and the coating has to suit the preparation. Losing material during processing is the most expensive slide problem there is, because the specimen is usually irreplaceable.
4. **Immersion medium matched to the objective**: Immersion objectives are designed around a medium of a specific refractive index and dispersion. Using a different medium, or a general-purpose oil on an objective designed for another, costs exactly the resolution the objective was bought for. Match the medium to the objective, not to what is on the shelf.
5. **Cleaning, which is a consumable too**: Lens tissue, the correct solvent and a routine matter more than any accessory. Residual oil on a high-power objective softens every image taken afterwards, and dried oil is a service job. Budget the cleaning materials and write the routine down beside the instrument.

**Does coverslip thickness on glass slides for microscopes really matter?**

For high-power dry objectives, yes. They are corrected for one thickness, and departing from it degrades the image in a way that is easily mistaken for a focus fault. Buy the graded thickness the objectives specify and standardize on it.

**Can I use any immersion oil?**

No. Immersion objectives are designed around a medium of a specific refractive index, and using another costs the resolution the objective was bought for. Match the medium to the objective and clean it off after every session.

**When is an isolation table worth buying?**

When high-power, long-exposure or probe-based work shows a soft or doubled image that improves outside working hours. Identify the frequencies involved first: relocating the instrument away from a pump, a corridor or a centrifuge sometimes fixes it for nothing.

## Microscope imaging software and scanning

Software outlives the camera and often the instrument, and it is where the long-term cost of an imaging system actually sits. Two systems with similar pictures can differ completely in whether the acquisition settings are stored, whether the files can be read by anything else, and whether an image can be traced back to the instrument and the operator that produced it. Those are the questions to ask first.

1. **What travels with the image**: Objective, illumination setting, exposure, filter set, calibration, timestamp and operator should be written into the file rather than into someone's notebook. An image without its acquisition settings cannot be compared with another image, which quietly ruins datasets collected over months.
2. **File format and whether anything else can read it**: A proprietary format is fine while the software runs and expensive afterwards. Ask whether the system writes or exports an open format without losing metadata, because that single answer decides whether an archive stays usable when the software is replaced.
3. **Licensing, seats and offline analysis**: Acquisition licenses are often tied to the instrument and analysis licenses to a seat, and analysis away from the instrument is what a busy laboratory needs most. Establish how many people can work at once, on what machines, and what happens when a license lapses.
4. **Measurement and calibration discipline**: Any measured value needs a calibration made at that magnification on that instrument, stored with the image and re-checked on a schedule. Software that lets a scale bar be typed in rather than derived from a calibration is producing decoration, not measurement.
5. **Storage, backup and who pays for it**: Imaging produces data quickly, and whole-slide and time-lapse work produce it very quickly. Decide where images live, how they are backed up, who pays after the project ends and how they are found again in three years, before the first dataset rather than after it.

**What should imaging software store with every image?**

Objective, illumination and exposure settings, filter set, calibration, timestamp and operator. Without those an image cannot be compared with another, and a dataset collected over months becomes a folder of pictures rather than a result.

**Is a confocal instrument better than a widefield fluorescence microscopy stand?**

Only for thick specimens that need optical sectioning. For thin specimens a widefield stand with a good camera collects more light in less time and bleaches the specimen less, which is why many laboratories keep both.

**Should I buy whole-slide scanning?**

If reading, sharing and archiving slides is a bottleneck, yes, because it changes the workflow rather than the picture. Evaluate it on a realistic mix of difficult preparations and price the storage, because both decide whether it saves time.

## Buying a centrifuge new, used or refurbished

A centrifuge is one of the better second-hand buys in a laboratory and one of the easier ones to get wrong, because the part that carries the risk and most of the value is not the instrument. Drives are durable and serviceable; rotors have a finite life, a documented history and a failure mode that wrecks the machine around them. Everything worth asking about a used instrument is really a question about its rotors and its records.

1. **Which rotors are included, and what life is left**: A rotor has a rated number of runs or years at full speed, and a responsible seller keeps a log. Ask for the rotor part numbers, their age, the log if one exists and the maker's derating or retirement policy. A machine with no rotor, or with a rotor at the end of its life, is a drive with a large purchase still attached.
2. **The service history and who did it**: Ask when it was last serviced, by whom, what was replaced and whether the maker will still support the model. Support status decides the instrument's remaining life more than its age does, and a discontinued model without parts is worth what its rotors are worth.
3. **What a refurbishment included**: Refurbished means different things to different sellers. The useful questions are whether the drive bearings were replaced, whether the lid interlock and imbalance detection were tested, whether the refrigeration was serviced and whether the instrument was run at full speed with a loaded rotor afterwards. Ask for that test to be recorded.
4. **Safety systems, tested rather than assumed**: The lid interlock, the imbalance cut-out and the rotor identification are the systems that stand between a failure and an injury. They are also the systems nobody checks on a machine that appears to work. Require evidence that each was tested, and re-test them at installation.
5. **The cost that arrives afterwards**: Count the rotors you will need, the adaptors for your vessels, a service contract or the cost of a visit, and the replacement of any consumable vessel. On a ten-year view these regularly exceed the purchase price, and a comparison that omits them is comparing the wrong number.

**Is a used centrifuge a safe purchase?**

It can be an excellent one, provided the rotor's age and run history are known, the safety interlocks are tested and the maker still supports the model. The rotor carries the risk and most of the value, so a machine sold without rotor records is an unknown.

**What should a refurbishment have included?**

Drive bearings checked or replaced, lid interlock and imbalance detection tested, refrigeration serviced where fitted, and a full-speed run with a loaded rotor afterwards, recorded. Ask for that record rather than the word.

**Does the quoted centrifuge machine price include a rotor?**

Often not, and that is the largest single variable. Ask which rotors are included by part number, what life is left on each, and what the adaptors for your own vessels cost, then compare.

**Where do a single use centrifuge and a vacuum centrifuge fit?**

Neither is a general purpose machine. A single use format exists so that a process step carries no cleaning validation between products, which is a manufacturing argument rather than a laboratory one. A vacuum concentrator spins to dry rather than to separate, and it is bought for removing solvent from small volumes; if that is the job, the specification to compare is the pumping speed and the chemical resistance, not the rotational speed.

## Benchtop autoclaves

A small chamber on a bench is the right answer for a great many laboratories and the wrong one for a few, and the difference is almost never the volume. It is the cycle: what the machine does with the air in the chamber, whether it can dry a wrapped load, and whether it can cool a liquid safely. Those three questions decide what a small machine is actually capable of.

1. **How the air leaves the chamber**: Steam cannot reach a surface that air is still occupying. The simplest machines displace air by letting steam push it out, which works for unwrapped solid items and not reliably for anything porous or hollow. Machines that actively remove air before admitting steam handle wrapped and hollow loads, and cost more.
2. **Whether the load has to come out dry**: A wrapped load that comes out wet is not sterile once handled, because moisture carries organisms through the wrap. Drying is a separate stage that needs a vacuum or a heated cycle, and a machine without one cannot process wrapped goods however good its sterilising phase is.
3. **Whether liquids are in scope**: Liquids need a slow, controlled cooling phase and a temperature probe to confirm the load rather than the chamber reached temperature. Many small machines offer a liquid cycle and some do not. If media preparation is the reason for the purchase, this is the first specification to confirm.
4. **Chamber shape against the real load**: Usable volume is what fits between the shelves with clearance for steam, and a cylindrical chamber wastes the corners of a rectangular tray. Measure the tallest and widest item and count the trays of a typical run before comparing litres.
5. **Water supply, drainage and the bench**: Small machines are usually filled by hand with treated water and drain into a reservoir that has to be emptied. Both are daily tasks and both are usually discovered after installation. Ask what water quality is required, how much a cycle uses and where the waste goes.

**Can tabletop autoclaves sterilise bagged waste?**

Generally not reliably. Bagged waste is dense, insulating and full of trapped air, and a displacement cycle will not reach temperature throughout it. The cycle will appear normal and the load will not be sterile.

**Do I need a vacuum cycle?**

If the load is wrapped, porous or hollow, or has to come out dry, yes. Displacement cycles suit unwrapped solid items and liquids where a liquid cycle is offered.

**How do I know the cycle worked?**

From an indicator placed in the hardest position within a representative load, at a stated frequency, with the result recorded. The chamber gauge tells you about the chamber, not about the load.

## Autoclave price and buying routes

The machine is the part of this purchase that is easiest to price and the smallest part of the cost. Steam, water treatment, drainage, ventilation, floor loading, commissioning, validation and a statutory inspection regime all arrive with it, and they are quoted by different people at different times. A comparison that stops at the machine is comparing the least variable number in the project.

1. **Machine, and what is bundled into it**: Establish whether the quotation includes the trays, baskets, trolley, printer or data logger, the loading equipment and the first service. These are frequently separate lines and frequently assumed, and they are the usual reason two quotations that looked level are not.
2. **Services the building has to bring**: Steam or an internal generator, treated water, a drain that will take the discharge, power, ventilation for the heat and a floor that will take the weight. Each is a line in someone else's budget and each has stopped an installation somewhere.
3. **Commissioning and validation**: Installation, operation and performance qualification with your own load types is what turns a machine into a process. It is a real cost with a real timetable, it is often not in the equipment quotation, and in a regulated setting it is not optional.
4. **The statutory regime for a pressure vessel**: A machine of any size is a pressure vessel with an inspection scheme, a competent person and a recurring examination. Ask what scheme applies, who writes it, what the examination costs and how often it falls due before the machine is chosen.
5. **Running cost over the life**: Water, treatment, energy, the service contract and the consumables are the number that decides which machine is cheaper, and they differ far more between designs than the purchase price does. Ask for water and energy per cycle for the cycles you will actually run.

**Why do two quotations differ so much?**

Because they usually describe different scopes. One may include trays, trolley, commissioning and a first service while the other prices the machine alone. Ask for a delivered, installed and commissioned price against your own load types.

**What does installation actually involve?**

Steam or a generator, treated water, a drain rated for the discharge, power, ventilation for the heat and floor loading, followed by commissioning and qualification with your own loads. Those belong on a drawing before the machine is ordered.

**Are refurbished autoclaves worth considering?**

Yes, with the same building requirements as a new one plus a history: chamber condition, door seal and gasket mechanism, controller support and the statutory examination records. Put validation into the comparison whichever route you take.

## Buying and authenticating a cell line

A cell line is a reagent whose identity can drift, be mistaken or be lost entirely, and unlike a chemical it cannot be checked by looking at it. The single decision that matters most is where it came from: a line from a repository arrives with a characterised identity, a passage number and a contamination status, and a line from a neighbouring laboratory arrives with a history nobody wrote down.

1. **Provenance you can cite**: A repository line has a catalog identifier, a documented origin and a published characterisation, and that identifier is what a methods section cites. A line received informally has none of those, and reconstructing them later is more expensive than buying the line was.
2. **Authentication, on arrival and on a schedule**: Short tandem repeat profiling confirms a human line is what it claims to be, against the repository's reference profile. Do it when the line arrives, before a bank is made, and periodically afterwards. Misidentified lines are common enough in the literature that funders and journals now ask for the evidence.
3. **Mycoplasma status, tested rather than assumed**: Mycoplasma contamination is invisible, common, and changes cell behavior in ways that look like biology. Test on arrival, quarantine until the result is known, and test the bank before it is used. A line that enters a shared incubator untested puts every other line in it at risk.
4. **A bank made early, at a low passage**: Expand and freeze a master and a working bank as soon as the line is authenticated and clean. Everything afterwards comes from the working bank at a recorded passage, which is what makes an experiment two years later comparable with one today.
5. **Terms of use, read before the order**: Lines carry conditions: research use only, restrictions on transfer to third parties, and sometimes obligations if the work becomes commercial. Read them at purchase rather than at the point a collaboration or a license is being negotiated.

**Why purchase cell lines from a repository rather than a colleague?**

Because the repository supplies a citable identifier, a characterised identity, a known passage and a contamination status. A line passed between laboratories carries a history nobody recorded, and reconstructing it costs more than the purchase would have.

**How often should a line be authenticated?**

On arrival, before a bank is made, and periodically afterwards, with the profile kept. Journals and funders increasingly ask for that evidence, and it is far easier to produce prospectively than to reconstruct.

**How many lines make a panel?**

Enough to span the properties your conclusion depends on, each chosen because it represents one of them. A panel assembled from what grows well is a convenience; a panel where each line stands for a stated property is an argument.

## Lung and breast cell lines

Lung and breast panels are the two best-characterised collections in common use, and both are chosen the same way: each line is included because it represents a molecular state the conclusion depends on. A panel picked for growth rate rather than genotype produces results that are reproducible within the laboratory and mean nothing outside it.

1. **Name the property each line stands for**: Driver mutation, tumour suppressor status, receptor expression and resistance mechanism are the properties that usually matter. Write each line's role in the panel down before ordering; if a line has no stated role it is there for convenience rather than for evidence.
2. **Include a sensitive and a resistant comparator**: Where the work concerns a targeted agent, the panel needs a line that responds and a line that does not, ideally differing by a known mechanism. Without both, a dose response is a description rather than a comparison.
3. **Include a non-tumour reference where the claim is selectivity**: Immortalised normal epithelial lines exist for exactly this and they are not normal tissue: they are immortalised, culture-adapted and sometimes transformed. They are a useful comparator when described accurately and a misleading one when described as normal cells.
4. **Match the species to the question**: Mouse lines exist so that a tumour can be grown in an immunocompetent animal of the same strain. That makes them the right choice for immune work and the wrong choice for anything about human genetics, and the strain background has to match the host or the experiment fails for a reason that has nothing to do with biology.
5. **Check receptor and marker status against the repository**: Receptor expression in long-passaged lines drifts and is reported inconsistently in the literature. Take the repository's characterisation as the reference and confirm the markers your conclusion depends on in your own hands rather than citing a paper from twenty years ago.

**Why include a resistant line beside a PC9 cell line?**

Because a dose response in sensitive lines alone describes the agent rather than testing a mechanism. A resistant comparator differing by a known mechanism turns the experiment into a comparison.

**Are immortalised epithelial lines normal cells?**

No. They have been altered to grow indefinitely, which changes the biology being compared. They are useful references when described accurately; where selectivity against normal tissue is the claim, primary cells are the stronger comparator.

**Can I use a mouse tumour line for human genetics?**

No. Those lines exist so a tumour can grow in an immunocompetent host of the same strain, which is right for immune work and wrong for human genetics. The strain has to match the host or the experiment fails for non-biological reasons.

**Is an immortalised bronchial line normal airway epithelium?**

No. It is immortalised and culture-adapted, which makes it a useful non-tumour comparator rather than normal tissue. Where selectivity against normal airway is the claim, primary cells are the stronger comparator.

## Blood and immune cell lines

Suspension lines are easier to grow than adherent ones and harder to use well, because most of them are bought for a behavior rather than for a phenotype: a line that differentiates on treatment, a line whose survival depends on one signal, a line that presents antigen. What the line does under a stimulus is the product, and that behavior drifts with passage more visibly than a growth rate does.

1. **Establish what the line is a model of**: A line that differentiates toward a mature phenotype on treatment models a process; a line that dies without a cytokine models a dependence; a line derived from a tumour models that tumour. These are different uses and a line is rarely good at more than one of them.
2. **Fix the differentiation protocol before the experiments**: Where the readout follows a differentiation step, the agent, the concentration and the duration are part of the model and change the result substantially. Fix them, record them, and treat a change in protocol as a change in the model rather than a detail.
3. **Watch the passage window**: Differentiation capacity and cytokine dependence both fade with continued passage, often before anything looks wrong. Work within a stated passage window from a banked stock and return to the bank rather than carrying a culture indefinitely.
4. **Keep the dependence in a dependence assay**: A line that survives only with one signal is a clean readout for whether that signal is active, and it stops being clean the moment a subpopulation adapts. Include the withdrawal control in every run, not only when the assay was set up.
5. **Know when the answer is a primary cell**: Some populations have no representative line at all, and work on them uses sorted primary cells with donor variability, limited numbers and a lifespan. Recognising that early avoids a search for a line that does not exist.

**Why do my differentiation results drift?**

Usually passage. Differentiation capacity fades before anything looks wrong, so work within a stated passage window from a banked stock and return to the bank rather than carrying a culture indefinitely.

**Is a THP1 cells monocytic line a substitute for primary macrophages?**

It is a tractable model of the process, not the mature cell. Use it to generate and screen hypotheses and test the important ones in primary cells, and state which protocol was used to differentiate it.

**What does a factor-dependent line measure?**

Whether a signal is active, because the cells die without it. Include the withdrawal control in every run, since subpopulations adapt and a line that has become independent reports nothing.

**Why are so many mouse immune experiments built on a few epitopes?**

Because the matching receptor-transgenic mice, tetramers and peptide reagents exist, which makes results comparable between laboratories. The trade is that affinity and immunogenicity are often far above anything physiological, so results transfer as mechanism rather than magnitude.

## Solid tumour cell lines

Outside the best-characterised tissues, a solid tumour line is usually chosen for one property: a pathway that is constitutively on, a hormone response that is present or lost, a driver mutation, or an ability to invade a matrix. Choosing on that property, and saying so, is what separates a model from a convenience, and it also decides which controls the experiment needs.

1. **Identify the single property the experiment turns on**: Hormone dependence, a pathway constitutively active through a known lesion, a driver mutation or an invasive phenotype. Name it, then find the line that carries it and the line that does not, because the pair is the experiment and either alone is a description.
2. **Use derivative pairs where they exist**: Many lines have a resistant or selected derivative from the same parent. A parent and derivative differ in far fewer ways than two unrelated lines, so a difference between them is much easier to attribute. Order both together and bank them together.
3. **Match the host for anything in vivo**: A human line needs an immunodeficient host and therefore cannot answer an immune question; a mouse line in a matching strain can. Deciding this before the line is chosen saves the discovery that the planned model cannot address the planned question.
4. **Confirm the lesion rather than citing it**: Mutation and expression status reported for a line decades ago is not always what your stock carries. Where the property is the reason for the choice, confirm it in your own stock once, at a known passage, and record it with the bank.
5. **Remember which lines are not tumour models**: Several familiar lines are neurobiology or differentiation models derived from other tissue entirely, and they are used for a behavior rather than for a cancer. Describing them as tumour lines misleads the reader about what the experiment was.

**Why use a parent and derivative pair?**

Because they differ in far fewer ways than two unrelated lines, so a difference between them can be attributed. Order and bank both together, or the comparison stops being controlled the first time one is replaced.

**Does the culture medium affect a 22RV1 cell line's hormone response?**

Considerably. Serum carries hormones, so stripped serum, a defined medium and a stated deprivation period are part of the method. Results in ordinary serum are not comparable with results in stripped serum.

**Why do invasion results stop reproducing?**

Most often because the matrix batch changed. Natural matrices vary substantially between lots, so record the matrix, its batch and its concentration with the line and treat a new lot as a change in the assay.

**How do I know which stock a bare designation refers to?**

Resolve it at the repository: look up the accession, read which names are recorded as synonyms and which as distinct substrains, and record the accession in your methods.

**What controls does a conditional deletion need?**

The recombinase alone, to show expressing it changes nothing, and the flanked allele alone, to show the modification is silent. Reporting the phenotype without both leaves the obvious question open.

## Tissue-specific human cell lines

For most human tissues the choice is between a primary cell that behaves like the tissue and runs out, and an immortalised line that grows indefinitely and has been changed to do so. Neither is correct in general. What decides it is whether the property you are studying survives immortalisation, and that is a question about your specific readout rather than about cell lines as a class.

1. **Ask whether the property survives immortalisation**: Immortalisation alters cell cycle control and frequently alters differentiation, barrier function and responsiveness. If your readout is one of those, the line may not show it. If your readout is a signalling event or a transport process that is preserved, the line is the practical choice.
2. **Account for donor variability**: Primary cells differ between donors, which is a nuisance in a mechanism experiment and the whole point in a study about population variation. Decide which of those you are doing, and where donors matter, use several and report them.
3. **Respect the tissue bed**: Cells from different vascular beds and different tissues differ substantially in marker expression, barrier tightness and response. A line from one bed is not a general model of that cell type, and the bed belongs in the description.
4. **Check what the line retains, in your own hands**: Suppliers state which markers and functions a line retains, usually at low passage under their conditions. Confirm the ones your conclusion depends on in your own culture, at the passage you use, before the line becomes the basis of a project.
5. **Consider a derived model where the question is genetic**: Where the question concerns a specific genotype, a differentiated stem-cell-derived model or a line edited at the locus of interest answers it more directly than any catalog line. Both are slower and more expensive, and both remove the argument about whether the model carries the variant.

**Primary cells or an immortalised line?**

It depends on whether your readout survives immortalisation. Cell cycle control, differentiation and barrier function commonly do not; many signalling and transport processes do. Decide on the specific readout rather than on the class of model.

**Is one endothelial cell line a general model?**

No. Cells from different vascular beds differ in markers, barrier tightness and response, so the bed belongs in the description and a line from one bed does not stand for the rest.

**When is an edited line worth the cost?**

When the conclusion depends on one genetic difference. An isogenic pair shares everything except the variable, which is the cleanest comparison available; buy one where it exists rather than making it.

## Petri dishes

A culture dish is chosen on three numbers: the growth area, the medium volume that area needs, and the cell yield it produces. Everything else, including the dimension in the catalog name, follows from those. A protocol moved between sizes without recalculating them is the commonest reason a confluent culture in one laboratory is a starving one in another.

1. **Work from growth area and yield**: The useful figures are the surface area of the dish and the number of cells it holds at confluence for your line. Write those down once for each format you use, and scaling a protocol becomes arithmetic rather than guesswork.
2. **Medium volume follows area, not dish depth**: Medium depth affects gas exchange, so the volume is set by the area rather than by how much the dish will hold. Too little dries and exhausts; too much slows oxygen transfer and changes the result. The supplier's recommended volume is a good starting point and worth checking against your own feeding schedule.
3. **Surface treatment is part of the product**: Untreated polymer is hydrophobic and suits suspension work or bacteriology. Tissue-culture treated surfaces are modified to let adherent cells attach, and specialized coatings go further. An adherent line put into an untreated dish simply will not attach, which is easy to mistake for a cell problem.
4. **Vented or not, and stacking**: Lids with molded ridges admit gas exchange and reduce condensation; tight lids suit anaerobic and long incubations. Dishes also have to stack stably in the incubator, and a stack that slides is a contamination route as well as a nuisance.
5. **Sterility and single use**: Polymer dishes arrive sterile and are used once. Glass is reusable, autoclavable and used where a solvent would attack polymer or where a long incubation or a high temperature is involved, at the cost of washing, wrapping and sterilising.

**Why will my cells not attach to a new petri dish?**

Check whether the dishes are tissue-culture treated. Untreated dishes sold for microbiology look identical and adherent cells will not attach to them. Store and label the two kinds separately.

**How much medium should a dish hold?**

Enough to cover the area without slowing gas exchange, set by growth area rather than dish depth. Start from the supplier's recommended volume and check it against your own feeding schedule.

**Is one large dish equivalent to several small ones?**

Not for uniformity. Edge and center conditions differ more in large dishes, and several smaller dishes also let one replicate be lost without losing the experiment. One large dish wins on cell yield per handling step.

**Is a petri dish holder worth having?**

For imaging and for dissection, yes: a holder stops the dish sliding on the stage and keeps it level, which matters as soon as anything is being measured rather than looked at. For storage and incubation a stack and a tray are enough, and the holders sold for that are mostly convenience.

## Single-use bags and containers

A single-use container is a polymer film with a set of ports, and almost every question worth asking about one is about the film: what it is made of, what it releases into the product, how it behaves at the temperature it will see, and who supplies it. The volume printed on the catalog page is the least informative number on it.

1. **The film, and what it releases**: Films are multi-layer laminates and each layer contributes. Extractables and leachables data for the specific film, under conditions like yours, is what tells you whether the container is compatible with the product. A supplier who cannot provide that data cannot support a regulated process.
2. **Qualification at the temperature it will see**: A film that is tough at room temperature may be brittle at cryogenic temperature, and a bag that survives freezing may fail when handled while frozen. Cryogenic containers are qualified for the whole cycle: freezing rate, storage, transport and thaw.
3. **Ports, tubing and how it connects**: Ports, tube sets and connectors are the difference between a closed process and an open one. Decide the connection scheme first, because a container with the wrong ports cannot be adapted and the assembly is usually the long lead item.
4. **Supply security and change control**: A single-use process depends on one supplier's film and assembly. Ask about second sources, notice periods for change, and what happens when a film is discontinued, because a film change is a process change that has to be assessed and sometimes requalified.
5. **Sterilisation and what it does to the film**: Assemblies are usually supplied irradiated, and the dose affects the film and its extractables profile. The validated dose range is part of the specification, and a request for a different dose is a request for a different product.

**What matters most about a single-use bag?**

The film: its composition, its extractables and leachables data under conditions like yours, and its qualification at the temperature it will see. Volume and price tell you the least.

**Why does a film change matter?**

Because the film is in contact with the product, so a change alters the extractables profile and is a process change to be assessed. Ask about second sources and notice periods before designing a container in.

**Is single-use always cheaper?**

No. It removes cleaning and changeover, which wins in multi-product and early-stage facilities. Above a certain scale and batch frequency, stainless steel with validated cleaning costs less per batch.

**How do I size a barbed fitting?**

By the tubing's inside diameter, not the port. Write the whole fluid path down as components with their sizes before ordering; mismatches are the usual reason a rig cannot be assembled when it is needed.

**Can I use braided tubing in a peristaltic pump?**

No. The braid is what stops it collapsing under pressure and it makes it unsuitable for a pump head. Rigs normally mix a pump-compatible section with braided runs elsewhere.

## Used laboratory equipment

Laboratory equipment holds value for a long time and turns over for reasons that have nothing to do with condition: a grant ends, a department closes, a company restructures. That makes the second-hand market genuinely good value and unusually opaque, because the seller often knows less about the instrument than the buyer needs to.

1. **Where it came from and what was in it**: An instrument from a laboratory that handled biological, radioactive or highly toxic material needs a decontamination declaration before anyone opens it. Reputable dealers provide one as a matter of course; a private sale frequently cannot, and that is a reason to decline rather than to negotiate.
2. **Whether it is still supported**: Parts availability, service support and, for anything computer-controlled, whether the software runs on a supported operating system decide the instrument's remaining life. An unsupported instrument can still be a bargain and it is a different purchase, and it should be priced as one.
3. **Seen working, on your own material where possible**: A demonstration on the seller's bench with a real sample is worth more than any description. Where that is impossible, as at auction, price the risk: assume a service, assume a consumable set and assume a qualification before first use.
4. **Moving it, and what that costs**: Decommissioning, rigging, transport, installation and recommissioning are real costs that frequently exceed the hammer price on large equipment. Ask who is responsible for removal and by when, because auction terms usually make that the buyer's problem on a short deadline.
5. **Statutory items need their paperwork**: Pressure vessels, lifting equipment and anything with a statutory examination scheme need their records to transfer with them. Without the history, a new scheme and an initial examination are required before use, which is a cost and a delay rather than a formality.

**What should I insist on before buying?**

A decontamination declaration, evidence of support and parts availability, and where possible seeing it work. At auction, assume a service, a consumable set and a qualification before first use, and price that in.

**What sells well when selling used lab equipment?**

Mechanically robust equipment with long lives and available parts: centrifuges, incubators, cabinets, baths and benching. Anything whose value is in software ages on a technology cycle and sells poorly.

**How do I sell my lab equipment for a good price?**

Clean it, gather the manuals and accessories, get a recent service record and a decontamination certificate, and show it running. Then choose the route by how quickly the space is needed: dealer, consignment, auction or direct sale.

## Food testing laboratories

Food testing is commissioned for three quite different reasons: to release a batch against a specification, to support a label claim, and to investigate a problem. Each needs a different scope, a different method and sometimes a different laboratory, and most disappointing reports come from a quotation written before anyone decided which of the three was being bought.

1. **Say which question the report has to answer**: Batch release tests against a written specification and produces a pass or fail. A label claim needs a method the regulator accepts and a result with an uncertainty. An investigation needs an open scope and a laboratory willing to follow evidence. A quotation cannot be compared until which of these is named.
2. **Check the schedule for your method and matrix**: Accreditation is granted method by method and matrix by matrix. A laboratory accredited for a pathogen in dairy may not be accredited for it in a dry powder, and a result outside the schedule is not an accredited result. Read the schedule rather than the certificate.
3. **Name the method, not just the analyte**: Different methods for the same analyte give different answers: a culture method and a molecular one detect different things, and a screening immunoassay and a confirmatory instrumental method report different numbers. Where a customer or a regulator specifies a method, that is the one that counts.
4. **Sampling decides the result**: A laboratory reports the sample it received. For a heterogeneous product, or for a contaminant that occurs in patches, the sampling plan is what connects the result to the batch, and a plan with too few increments produces a result that means nothing about the lot.
5. **Establish reporting limits and turnaround honestly**: Ask for the reporting limit for each analyte in your matrix, and for turnaround measured from sample receipt to report rather than from the start of the analysis. For a microbiological result the incubation sets a floor no laboratory can beat.

**Is a food testing lab accredited for my product?**

Not necessarily. Accreditation is method by method and matrix by matrix, so a laboratory accredited for a pathogen in one matrix may not be for another. Read the schedule and confirm your method and matrix appear on it.

**Can a rapid kit replace lab testing for food products?**

For screening, swab verification after cleaning and incoming checks, yes, and it is far cheaper. Confirm positives by a confirmatory method, and treat a negative as a screening result rather than a release, because processing changes proteins and matrices interfere.

**How is a shelf life established?**

By defining failure, storing the product at realistic conditions and measuring microbiological, chemical and sensory endpoints until one fails, which takes as long as the life claimed unless an accelerated design is justified. Challenge testing gives the stronger pathogen answer.

Get a shortlist: https://biobricks.org/quote/ | All service classes: https://biobricks.org/services/