sgRNA ordering: design, format and libraries

From design to a confirmed cut, with a synthetic guide or a gRNA plasmid

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 optimising 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.

Design tools rank, they do not predict

Scoring algorithms are useful for excluding obviously poor guides and for flagging off target risk, and they are weak at predicting which of several good candidates will cut well in your cells. Chromatin state, delivery and repair all intervene.

Treat the score as a filter and the bench as the test. Three guides screened empirically beats one guide chosen confidently.

What a library screen really costs

Representation is the constraint. Maintaining enough cells per guide at transduction, during selection, at sorting and at sequencing sets the scale of everything, and for larger libraries that scale is substantial in flasks, media and sequencing.

Calculate backwards from library size at the planning stage. Screens fail on representation far more often than on library quality.

Controls, in the order they matter

A non-targeting guide delivered and selected identically is the baseline, because delivery and selection themselves change cells. A guide against a gene with a known phenotype confirms the system works in your hands. More than one guide against the target is what separates an on-target effect from an off-target one.

For any conclusion that rests on the edit, a rescue with the coding sequence, or reversion, is the strongest control available. It is more work and it is the control that survives review when a phenotype is surprising.

Measuring the edit

A mismatch-cleavage assay gives a fast, cheap indication that editing happened and it quantifies poorly and misses small or homogeneous edits. Sequencing the amplicon gives the actual allele spectrum and is what a published editing efficiency should be based on.

Where a clone is the deliverable, sequencing both alleles and confirming the protein is absent are both needed: a frameshift can produce a truncated protein, and a clone carrying an in-frame deletion looks edited and is not a knockout.

Pooled and arrayed screens answer different questions

A pooled screen needs a readout that acts on the pool, such as survival or a sortable reporter, and it scales to whole genomes. Its output is guide abundance before and after selection, so the analysis has to handle guide-level variation, coverage and the statistics of a large multiple comparison.

An arrayed screen puts one guide per well and allows any readout, including imaging and multiplexed measurements, at a fraction of the scale. Where the phenotype cannot be selected for, arrayed is the only option, and its cost per gene is the constraint.

Other nucleases and why they persist

Alternative nucleases recognise different sequence motifs, which matters when no good guide site exists near the target position. Some leave different end structures, which changes the repair outcome and suits particular insertion strategies.

Older protein-guided platforms remain in use where a target has proved intractable and where an established regulatory path exists for a product. They are more work to build and they are not obsolete for those two reasons.

Testing regulatory sequence rather than editing it

A reporter library places thousands of candidate regulatory sequences upstream of a barcoded reporter and reads their activity in one experiment, which measures what a sequence does to transcription directly. It tests sequences out of their genomic context, which is both its power and its limitation.

An editing screen alters sequence in place and reads a phenotype, keeping the context and changing the genome. The two are complementary: the reporter assay nominates sequences with activity, and editing tests whether that activity matters where it actually sits.

sgrna crispr work, and what the guide has to be

A single guide RNA joins the targeting spacer and the scaffold in one molecule, and three things decide whether it works: the spacer's twenty nucleotides chosen for on-target activity and checked for close matches elsewhere, the PAM immediately downstream in the genome rather than in the guide, and the form it is delivered in, synthetic with end modifications for ribonucleoprotein work, or transcribed from a plasmid or a vector for a screen. Two or three guides per target are ordered because activity varies.

A non targeting control grna and what it controls for

A non targeting control grna matches the library's chemistry and delivery while binding nothing in the genome, which is what separates the effect of editing from the effect of the delivery, the selection and the nuclease. A scrambled sequence still has to be checked against the genome, because a control that cuts somewhere is worse than no control at all.

An oligo dt primer and when it is the wrong choice

An oligo dt primer anneals to the polyadenylated tail, so it biases reverse transcription toward the three prime end and misses non polyadenylated RNA entirely, including most bacterial transcripts and many non coding species. Random primers or a gene specific primer are the alternatives, and the choice belongs in the method because it changes what the library or the assay can see.

A lentiviral library and how it is delivered

A lentiviral library delivers one construct per cell at a low multiplicity so a phenotype can be traced back to a single perturbation, which is why titre and representation are measured before a screen rather than after it. Coverage is the number of cells per construct, and a screen that cannot state it cannot be read.

vector vs plasmid, and plasmid vs vector

In vector vs plasmid the words describe a role and a form: a plasmid is a circular DNA molecule, and a vector is whatever carries a sequence into a host, which may be a plasmid, a virus or a transposon. Written the other way round as plasmid vs vector the question is the same, and the answer is that most plasmids in a laboratory are vectors and not every vector is a plasmid.

Common questions

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.

Browse by service class

• Compare Gene synthesis on advertised prices