Exploring lead compounds with potent inhibition from hit compounds is one of the challenging tasks in drug discovery. Free energy perturbation (FEP) calculations have attracted attention as a lead optimization method that can predict binding affinities with extremely high accuracy [1]. However, FEP calculation requires huge computational time compared to other quantitative structure-activity relationships (QSAR) and molecular docking. Therefore, it is essential to explore derivatives in the possible chemical space more efficiently with fewer trials to take advantage of FEP for lead optimization.
Recently, active learning-based lead optimization based on relative binding free energy calculations was recently proposed [2]. Still, the effectiveness of this approach in practice has not been comprehensively investigated. This study aims to establish a workflow for FEP calculations based on active learning to explore new compounds with desirable binding free energies in fewer trials.
各種検索
サポート
ヘルプ