Molecular optimization by capturing chemist’s intuition using deep neural networks

• DOI: 10.1186/s13321-021-00497-0
• Publication Date: 2021-03-20T12:02:54Z
• Authors: Jiazhen He; Huifang You; Emil Sandström; Eva Nittinger; Esben Jannik Bjerrum; Christian Tyrchan; Werngard Czechtizky; Ola Engkvist
• Keywords: Molecular graph; Intuition; Molecular machine

Abstract A main challenge in drug discovery is finding molecules with a desirable balance of multiple properties. Here, we focus on the task molecular optimization, where goal to optimize given starting molecule towards This can be framed as machine translation problem natural language processing, our case, translated into optimized properties based SMILES representation. Typically, chemists would use their intuition suggest chemical transformations for being optimized. widely used strategy concept matched pairs two differ by single transformation. We seek capture chemist’s from using models. Specifically, sequence-to-sequence model attention mechanism, and Transformer are employed generate As proof concept, three ADMET simultaneously: logD , solubility clearance which important drug. Since often vary project project, user-specified property changes incorporated input an additional condition together Thus, models guided satisfying Additionally, compare representation, graph-to-graph HierG2G, has shown state-of-the-art performance optimization. Our results show that more making small modifications molecules, intuitive chemists. further enrichment diverse achieved ensemble