Structure-based lead optimization is an open challenge in drug discovery, which is still largely driven by hypotheses and depends on the experience of medicinal chemists. We here propose a pairwise binding comparison network (PBCNet) based on physics-informed graph attention mechanism, specifically tailored for ranking relative binding affinity among congeneric ligands. Benchmarking on two held-out sets (provided by Schrödinger, Inc. and Merck KGaA) containing over 460 ligands and 16 targets, PBCNet demonstrated significant advantages in terms of both prediction accuracy and computational efficiency. Equipped with a fine-tuning operation, the performance of PBCNet reaches that of Schrödinger's FEP+, which is much more computationally intensive and requires significant expert intervention. A further simulation-based experiment showed that active learning-optimized PBCNet may accelerate lead optimization campaigns by 30%. Finally, for the convenience of users, a web service (https://pbcnet.alphama.com.cn/index) for PBCNet is established to facilitate complex relative binding affinity prediction through an easy-to-operate graphical interface.
Supplementary information. Some illustrative tables and images are provided in this document.