Phage display antibody library: design, panning and outputs

Running a campaign

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 recognise 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.

Phage display library construction: selection finds what you presented

The most reliable rule in display is that the campaign answers the question the selection conditions asked. Immobilisation chemistry, buffer, competitor and washing stringency together define that question, and changing any of them changes the binders that emerge.

Write the selection conditions down as a hypothesis about the molecule you want, and check them against the assay the leads will face. Divergence between the two is where campaigns quietly go wrong.

What happens after the hit

A binder becomes useful only once it is reformatted, expressed, purified and characterised for affinity, specificity, stability and developability. That work is a substantial project of its own and is frequently omitted from a discovery budget.

Plan it at the start, including the conjugation or fusion format the end use requires, so that leads are triaged against the properties that will actually matter rather than against affinity alone.

Phage display for antibody discovery, and what a library cannot give you

Display selects binders from a library by panning against the target, which is fast, animal free and able to find antibodies against things an animal tolerates poorly. What it gives is a SEQUENCE and a binder, not a validated reagent: affinity is often modest before maturation, and the selection conditions shape what is found, so panning in a detergent finds binders that work in a detergent.

The practical consequence is that the campaign is designed around the counter selection as much as the selection: what you deplete against decides specificity more than the number of rounds does. Plan the off target panning and the screening assay before the first round, because a hundred clones that all bind the tag are the usual outcome of a campaign that skipped it.

A vhh library, and why single domains are displayed

A VHH is the variable domain of a camelid heavy chain antibody: a single fifteen kilodalton domain that folds and binds on its own, which makes it ideal for display because one gene fragment encodes the whole binder and there is no light chain to pair. Libraries are built from an immunised animal's repertoire, which gives affinity-matured binders, or synthetically from a framework with randomised loops. The long CDR3 reaches into clefts, which is why they find enzyme active sites and receptor pockets that conventional antibodies miss.

A per2 antibody and a circadian protein

PER2 abundance and phosphorylation cycle over the day, so a per2 antibody reports a time point and a figure without the collection time in the light cycle is uninterpretable. The protein is large and heavily phosphorylated, which shifts its apparent mass, and a synchronised culture with a stated hour after the synchronising stimulus is the only comparable sample.

A gad1 antibody and the other glutamate decarboxylase

GAD1 encodes the isoform usually called GAD67, so a gad1 antibody and a gad67 antibody are the same target under two conventions, and specificity against the other gene's product is what an inhibitory neuron claim rests on. It is cytoplasmic and fills somata and processes, which is how the pattern is told from the vesicle associated isoform.

An h3.3 antibody and a variant that differs by a few residues

Histone H3.3 differs from the canonical histone by a handful of amino acids, so an h3.3 antibody has to be raised against that divergent region and its specificity against the canonical protein is the only thing that makes it useful. The variant marks actively transcribed regions, and chromatin extraction rather than a whole lysate is the preparation.

A fosb antibody and the truncated isoform

FOSB is expressed as the full protein and as a stable truncated isoform that accumulates with repeated stimulation, and the two have different half lives, so a fosb antibody has to detect both and the figure has to name which changed. It is an immediate early product, so an unstimulated control and a stated time after stimulation are the controls.

Common questions

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 programmes 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 licence terms. A handful of clones with affinity values is not a campaign record.

Browse by service class

• Compare Gene synthesis on advertised prices