Auto in silico ligand directing evolution (AILDE) to facilitate the rapid and efficient discovery of drug lead

Abstract Drug discovery has become increasingly difficult because of the high costs and risks associated with obtaining drug lead. Because a large number of analogs need to be synthesized and validated, the hit-to-lead (H2L) stage is a rate-limiting step and thus becomes a bottleneck in drug discovery. Hence, medicinal scientists have developed various rational design approaches to address the challenge in the H2L stage and have reported many fantastic applications of these strategies. However, the rational design of drug leads in the H2L stage remains a formidable challenge due to unreachable chemical space, model inapplicability, and low efficiency in the decision-making with regard to the lead. We therefore developed auto in silico ligand directing evolution (AILDE) as an efficient and general approach for the rapid identification of drug leads in accessible chemical space. The deduction and decision-making with regard to the leads, as well as the applicability of AILDE, has been rigorously validated on dozens of drug targets. We also used AILDE to discover potent drug leads for mesenchymal-epithelial transition factor (c-Met), a kinase target implicated in cancer, by synthesizing only eight compounds. The identified drug leads have advanced to preclinical research. There have been no c-Met kinase drug leads discovered by highly efficient optimization strategies involving computational design, synthesis, in vitro and in vivo assays, and cocrystallization validation. We also developed a web service ( http://chemyang.ccnu.edu.cn/ccb/server/AILDE ) aiming to provide broad access to this platform. AILDE robustly identifies new leads for drug discovery by greatly reducing the chemical optimization burden.