Prospective evaluation and success of a machine learning hit-to-lead drug development program against phosphatidylinositol 3-kinase α

Abstract

As a result of the rapidly increasing cost of drug development, efficient methods for early identification of compounds with a high probability of clinical success are needed. Herein, we describe a cheminformatics protocol which dramatically increases quality candidate identification and should reduce the attrition rate of compounds entering the clinic, increasing the cost-effectiveness of drug development. Against the oncology target phosphatidylinositol 3-kinase α, all five compounds synthesized from the protocol were found to have low nanomolar activity. We therefore propose that our protocol can be used as a tool for reducing the synthetic burden required for hit-to-lead optimization.