Computational discovery of RSV Pre-F inhibitors via reinforcement learning-driven ab initio design from natural...

Respiratory syncytial virus (RSV) is a leading cause of lower respiratory tract infections in children, yet effective treatment options remain limited. In this study, we integrated natural compounds with a fragment-based reinforcement learning model and proposed a generative molecular design approach based on a natural compound fragment library to design RSV inhibitors by matching the 3D structure of the pre-fusion F protein (Pre-F). Based on physicochemical properties and docking scores, five candidate compounds were selected. Molecular dynamics simulations and binding free energy calculations confirmed their stable interactions with Pre-F. Further analyses using principal component analysis (PCA), free energy landscapes (FEL), and quantum chemical methods such as electrostatic potential (ESP) and conceptual DFT (CDFT) revealed favorable electronic and structural characteristics. These findings highlight the potential of our natural fragment-based design strategy for discovering RSV inhibitors.