Discovery of MAO-B Inhibitor with Machine Learning, Topomer CoMFA, Molecular Docking and Multi-Spectroscopy Approaches
Linfeng Zheng, Xiangyang Qin, Jiao Wang, Mengying Zhang, Quanlin An, Jinzhi Xu, Xiaosheng Qu, Xin Cao, Bing Niu
Abstract
Alzheimer’s disease (AD) is the most common type of dementia and is a serious disruption to normal life. Monoamine oxidase-B (MAO-B) is an important target for the treatment of AD. In this study, machine learning approaches were applied to investigate the identification model of MAO-B inhibitors. The results showed that the identification model for MAO-B inhibitors with K-nearest neighbor(KNN) algorithm had a prediction accuracy of 94.1% and 88.0% for the 10-fold cross-validation test and the independent test set, respectively. Secondly, a quantitative activity prediction model for MAO-B was investigated with the Topomer CoMFA model. Two separate cutting mode approaches were used to predict the activity of MAO-B inhibitors. The results showed that the cut model with q2 = 0.612 (cross-validated correlation coefficient) and r2 = 0.824 (non-cross-validated correlation coefficient) were determined for the training and test sets, respectively. In addition, molecular docking was employed to analyze the interaction between MAO-B and inhibitors. Finally, based on our proposed prediction model, 1-(4-hydroxyphenyl)-3-(2,4,6-trimethoxyphenyl)propan-1-one (LB) was predicted as a potential MAO-B inhibitor and was validated by a multi-spectroscopic approach including fluorescence spectra and ultraviolet spectrophotometry.