What it does
Epitope Scout takes a target structure (PDB or mmCIF, or a PDB ID fetched directly from RCSB) and returns a ranked list of surface epitope patches suitable for de novo binder design. Each patch is scored on a 0 to 1 scale combining hydrophilicity, sequence conservation, pocket geometry, and surface accessibility, with the top candidates visualised in 3D and exported as a residue-list CSV ready to drop into RFdiffusion or BindCraft as hotspot constraints.
The default workflow: select a target chain, wait for the scoring pass, review the top three patches in the embedded Mol* viewer, and export. A typical run on a single-chain antigen of 200 to 400 residues completes in well under a minute on CPU.
What it is not
Epitope Scout does not predict binder sequences, and it does not replace a structural biologist’s judgment on druggability. It replaces the manual PyMOL inspection step at the front of a binder campaign with a consistent, reproducible scoring pass, so that hotspot selection is driven by the same criteria every time rather than by who happens to be inspecting the structure that week.
Methodology
The five-component composite draws on the protein-protein interaction literature on surface druggability. Weights: hydrophobic exposure 30 percent, structural order 20 percent, rigidity 20 percent, surface accessibility 15 percent, hot spot density 15 percent. Rigidity uses crystallographic B-factors when available and AlphaFold pLDDT otherwise. Hot spot density counts canonical PPI interface residues (Trp, Tyr, Arg, Phe) within each patch. Full scoring details are documented on the tool itself and in the linked methodology page.
Non-scoring quality flags surface separately: loop-only anchors, all-polar patches, glycan proximity, and high local flexibility. These are displayed next to each ranked patch so you can weigh them against the score.
When to use it
Run Epitope Scout before committing compute to a de novo binder campaign. Hotspot selection is the single most consequential decision in a RFdiffusion or BindCraft run, and is the most common failure mode: generation converges on buried residues, disordered loops, or glycosylated sites and burns GPU hours on scaffolds that were never going to work. A 30 second scoring pass before launch catches most of these cases.
Also useful for: evaluating multiple candidate epitopes on a novel target, sanity-checking a chosen hotspot against known binders, and preparing residue selections for a wet-lab binder screen.
Pricing
Free tier includes 3 runs per rolling 30 day window, full scoring, and CSV export. Scout Pro at 49 USD per month removes the run limit and retains run history across projects. No credit card required for the free tier.