Data-Driven Design of Diagnostic Kits and Therapeutic Peptides

Abstract

Learning from experimentally determined interacting secondary structural motifs, we compiled a database to facilitate a data-driven design of therapeutic peptides (TPs). 1.7 million helical peptides (HPs) in >130 thousand proteins are extracted along with their interacting partners from the protein data bank (PDB). The sequences of the HPs are developed into a searchable database (TP-DB) by creating indices that map specific peptide patterns to locations of matched HPs in the TP-DB. Leveraging TP-DB to search for a potent membrane-insertion pattern WXXWXXW, established by our microsecond-long MD simulations, we found a positively charged HP that matches the pattern has a commensurate minimal inhibitory concentration (MIC) against Candida albicans (fungus) as compared to previously characterized homologs. With identifying peptides containing the affinity determinant motifs DYKXX[DE] of FLAG-tag within pathogenic proteins, which PHI-BLAST failed to find, we successfully discovered Helicobacter pylori neutrophil-activating protein (HP-NAP), a virulence factor of H. pylori, to contain a stretch of sequence DYKYLE that can be recognized by the anti-FLAG M2 antibody. By doing so, we repurposed a purification-tag-specific antibody into a diagnostic kit for pathogen’s virulence factors. Taken together, we believe that TP-DB and its pattern-based search engine provide a new opportunity for a (secondary-)structure-based design of peptide drugs and diagnostic kits for pathogens without inferring evolutionary homology between sequences sharing the same pattern. TP-DB is made available at http://dyn.life.nthu.edu.tw/design/