DeepMetab: a comprehensive and mechanistically informed graph learning framework for end-to-end drug metabolism prediction

DeepMetab is the first comprehensive, mechanistically informed deep graph learning framework for end-to-end prediction of CYP450-mediated drug metabolism. It uniquely integrates three essential tasks — substrate profiling, site-of-metabolism (SOM) localization, and metabolite generation — within a unified multi-task architecture, using a dual-labeling strategy that captures both atom- and bond-level reactivity and infusing quantum-informed and topological descriptors into a GNN backbone. A curated knowledge base of expert-derived reaction rules enforces mechanistic consistency during metabolite synthesis. DeepMetab consistently outperforms existing models across nine major CYP isoforms in all three tasks and generalizes to 18 recently FDA-approved drugs, achieving 100% TOP-2 accuracy for SOM prediction.