An open hub for AI-generated peptide candidates. Every sequence gets a card. Every card ships with a recipe. Every fork is traceable.
PeptideModel is an open research database for AI-generated peptide candidates. Every entry follows a five-level evidence pipeline — designed, computed, reproduced, synthesized, and bioassayed — that distinguishes AI proposals from experimentally validated sequences. Each card records the full prediction recipe (model, version, hardware, seeds, command) for independent reproducibility. All data is CC-BY-SA 4.0 and accessible via a free public API at api.peptidemodel.com. The platform currently indexes 1739 sequences across 64 biological targets.
The grid of family cards becomes a body-map on-ramp. Pick a system — by tab, by the figure, or by the index — and the panel resolves to that region's targets, ranked by activity, one click from the full list.
view anterior · male · icon-led systems · mapped rig
Every card carries a five-level status. Anyone can design; computation and reproduction require a recipe and attached artifacts; synthesis requires an LC-MS spectrum; assay requires a protocol and raw data. The funnel is steep because the schema is strict — and that's the point.