PeptiBase: a novel source for peptide-based drug development

Peptides are important molecules with diverse biological functions and medical uses. Databases describing peptide binding profiles to their targets1 are commonly employed for peptide-docking benchmarks. Currently, this task does not consider peptides’ affinity or decoy-peptides. Aiming to overcome these drawbacks, we assembled a dataset of peptide-proteins complexes from BindingMOAD,2 whose pKi ranges from 1.3 to 10.2 (Plot 1). Lys, Asn, Ile, and Thr residues are more frequent in peptides of the dataset than expected from natural frequency3 (Plot 2). Similarity searches using the peptides from our database as templates afforded ligands with high (actives) and low (decoys) affinity for their targets. Docking of active and decoy peptides with docking engines reveals Vina has poorer performance than Dockthor (Plot 3). Further studies are required to evaluate affinity’s impact on the docking success rates, but current results suggest that our database is suitable to benchmark docking engines.