Medicinal Chemistry and Drug Design

Biography

Biography: Orazio Nicolotti

Abstract

Pairing novel compounds to specific drug targets or repositioning old drugs to apparently unrelated diseases is a fascinating and challenging theme of rational drug discovery. In this scenario, we developed an easy-to-run in silico tool for drug target and quantitative bioactivity prediction implementing a multi-fingerprint similarity search algorithm, whose acronym is MuSSeL. Predictions were derived by exploiting a large collection of highly curated experimental bioactivity data available from ChEMBL (version 22.1) and combining results based on similarity search screening employing 13 different molecular fingerprints. Noteworthy, the occurrence of potential activity cliffs was also explicitly accounted in MuSSeL. The last updated version of ChEMBL (version 23) was employed for tuning algorithm parameters and three randomly built external sets were used for measuring model performances. Interestingly, the prospective use of MuSSeL was challenged predicting five real-life bioactive compounds taken from articles just published in Journal of Medicinal Chemistry whose structures and related bioactivity data were not covered in ChEMBL yet. Our approach demonstrated to return a more informed interpretation of drug target and bioactivity prediction and could be of valuable help to assist and speed up early stages of drug discovery programs.