Epitope binning
Epitope binning sorts a panel of binders by whether they block each other. Run pairwise, it produces a competition matrix and, from it, groups of antibodies that behave as though they share a site. It is the cheapest way to find out whether a panel has diversity, and it is routinely over-interpreted.
What a bin is
A bin is a set of binders that compete with each other. Competition means their binding is mutually exclusive on that antigen under those conditions. It does not necessarily mean they touch the same residues. Two antibodies can compete because their epitopes overlap, because they are close enough to clash sterically, or because one of them changes the antigen's conformation when it binds.
That distinction matters when the program depends on the epitope, for instance when the goal is to block a specific receptor interaction. Binning tells you the panel is diverse. Mapping tells you where a binder sits.
The two common formats
Premix, or in-solution competition. Antigen is preincubated with a saturating amount of one antibody, then flowed over a second antibody captured on a surface. Simple to interpret and it needs more antigen.
Sandwich, or classical. One antibody is captured, antigen is bound, then the second antibody is added to see whether it can still bind. Efficient with antigen and sensitive to orientation effects.
Both are usually run on SPR or BLI, and competition ELISA is a perfectly good low-throughput version when instrument time is scarce.
Controls that decide whether the matrix means anything
- Self-competition. Every antibody must block itself. An antibody that does not is not saturating, and its whole row and column are unreliable.
- Saturation check. Confirm that the first antibody is present in excess and that the antigen is fully occupied before the second is added.
- A known pair. Where a reference antibody with a known epitope exists, include it. It anchors the map to something real.
- Format consistency. Bivalent and monovalent forms behave differently in competition. Run the panel in one format.
- Antigen quality. Aggregated antigen produces apparent non-competition, because a second antibody can bind a second copy in the same aggregate.
Reading the matrix
Competition matrices are rarely symmetric, and the asymmetry is data. A pair where A blocks B but B does not block A usually indicates a size or accessibility difference, or a conformational effect rather than an overlap.
Cluster the matrix rather than eyeballing it, then look at the clusters against everything else you know: neutralization, species cross-reactivity and any structural prediction. Bins that line up with function are the ones worth carrying forward.
When you need real mapping
If the answer has to be residue-level, binning is not enough. Hydrogen-deuterium exchange mass spectrometry, alanine scanning, peptide arrays for linear epitopes, and structure determination by crystallography or cryo-EM all give different resolutions at different costs. A predicted epitope from co-folding is a hypothesis to test with one of these, not a substitute for them.
How we run it
We bin panels by competition with the self-competition and saturation controls in place, report bins as competition groups rather than as epitopes, and escalate to a mapping method when the program depends on where a binder sits.