QBMG: quasi-biogenic molecule generator with deep recurrent neural network

• Published in: Journal of cheminformatics Vol. 11; no. 1; pp. 5 - 12
• Main Authors: Zheng, Shuangjia, Yan, Xin, Gu, Qiong, Yang, Yuedong, Du, Yunfei, Lu, Yutong, Xu, Jun
• Format: Journal Article
• Language: English
• Published: Cham Springer International Publishing 17.01.2019; BioMed Central Ltd; Springer Nature B.V; BMC
• Subjects: Artificial neural networks; Backup software; Chemical properties; Chemistry; Chemistry and Materials Science; Computational Biology/Bioinformatics; Computer Applications in Chemistry; Deep learning; Digital libraries; Documentation and Information in Chemistry; Drug discovery; Natural product; Natural products; Neural networks; Novels; Optimization; Organic chemistry; Recurrent neural networks; Research Article; Technology; Technology transfer; Theoretical and Computational Chemistry; Training; Transfer learning; Virtual library; Deep learning; Natural product; Recurrent neural networks; Virtual library
• ISSN: 1758-2946; 1758-2946
• DOI: 10.1186/s13321-019-0328-9

Biogenic compounds are important materials for drug discovery and chemical biology. In this work, we report a quasi-biogenic molecule generator (QBMG) to compose virtual quasi-biogenic compound libraries by means of gated recurrent unit recurrent neural networks. The library includes stereo-chemical properties, which are crucial features of natural products. QMBG can reproduce the property distribution of the underlying training set, while being able to generate realistic, novel molecules outside of the training set. Furthermore, these compounds are associated with known bioactivities. A focused compound library based on a given chemotype/scaffold can also be generated by this approach combining transfer learning technology. This approach can be used to generate virtual compound libraries for pharmaceutical lead identification and optimization.