Electron-density informed effective and reliable de novo molecular design and lead optimization with ED2Mol

Abstract

Generative drug design opens new avenues for discovering novel compounds within the vast chemical space rather than conventional screening against limited compound libraries. However, the practical utility of the generated molecules is frequently constrained, as many designs prioritize a narrow range of pharmacological properties while neglecting physical reliability, which hinders the success rate of subsequent wet-lab evaluations. To address this, we propose ED2Mol, a deep learning-based approach that leverages fundamental electron density information to improve de novo molecular generation and lead optimization. The extensive evaluations across multiple benchmarks demonstrate that ED2Mol surpasses existing methods in terms of generation success rate and >97% physical reliability. It also facilitates automated lead optimization that is not fully implemented by other methods using fragment-based strategies. Furthermore, ED2Mol exhibits generalizability to more challenging, unseen allosteric pocket benchmarks, attaining consistent performance in both de novo molecule generation and lead optimization. More importantly, ED2Mol has been applied to various real-world essential targets, successfully identifying wet-lab validated bioactive compounds, ranging from FGFR3 orthosteric inhibitors to CDC42 allosteric inhibitors and GCK allosteric activators. The directly generated binding modes of these compounds with target proteins are close to predictions through molecular docking and further validated via the X-ray co-crystal structure. All these results highlight ED2Mols potential as a useful tool in realistic drug design with enhanced effectiveness, physical reliability, and practical applicability.