Exploration of peptides that fit into the thermally vibrating active site of cathepsin K protease by alternating artificial intelligence and molecular simulation

Abstract

Eighteen tripeptides that fit into the thermally vibrating active site of cathepsin K were discovered by alternating artificial intelligence and molecular simulation. The 18 tripeptides fit the active site better than the cysteine protease inhibitor E64, and a better inhibitor of cathepsin K could be designed considering these tripeptides. Among the 18 tripeptides, Phe–Arg–Asp and Tyr–Arg–Asp fit the active site the best and their structural similarity should be considered in the design process. Interesting factors emerged from the structure of the decision tree, and its structural information will guide exploration of potential inhibitor molecules for proteases.