ISSN print edition: 0366-6352 ISSN electronic edition: 1336-9075 Registr. No.: MK SR 9/7 Published monthly

In silico prediction of GLP-1R agonists using machine learning approach

Zheng-Kun Kuang, Xi-Yao Cheng, Zi-Xin Yang, Yu-Xi Guo, Yong-Qi Huang, and Zheng-Ding Su

Key Laboratory of Industrial Fermentation (Ministry of Education), Hubei Key Laboratory of Industrial Microbiology, National 111 Center for Cellular Regulation and Molecular Pharmaceutics, Hubei University of Technology, Wuhan, China

E-mail: zhengkunkuang@hbut.edu.cn

Received: 19 October 2020  Accepted: 9 March 2021

Abstract:

Glucagon-like peptide 1 receptor (GLP-1R) is a well-known drug target for the treatment of type 2 diabetes mellitus (T2DM). However, the currently marketed peptidyl GLP-1R agonist drugs are restricted by the requirement of injection. Hence, there is a continued need to develop orally bioavailable small molecule GLP-1R agonist drugs that could be beneficial for the treatment of T2DM. In this study, we report a new strategy to predict small molecule GLP-1R agonists with machine learning approach. Several regression and classification models were built based on support vector machine algorithm and diverse compounds with molecular properties and structural fingerprints as descriptors. For regression models, the ten-fold cross-validation squared correlation coefficient (q², for training sets) and determination coefficient (r², for test sets) of the optimized models were greater than 0.6, respectively. For classification models, the overall predictive accuracies were around or over 90% (for test sets). The results demonstrated that these reliable models could be used to identify highly active agonists for the purpose of virtual screening. The important properties and structural fragments for GLP-1R agonists derived from these models can be used for the novel GLP-1R agonist scaffold design.

Keywords: Glucagon-like peptide 1 receptor; GLP-1R agonists; Support vector machine; Virtual screening; Cheminformatics

Full paper is available at www.springerlink.com.

DOI: 10.1007/s11696-021-01600-3

Chemical Papers 75 (7) 3587–3598 (2021)