Virtual Screening of Acetylcholinesterase Inhibitors Based on Machine Learning Combined with Molecule Docking Methods

Abstract

Objective: The aim of this study was to screen for compounds with relatively high inhibitory activity on acetylcholinesterase. Methods: Classification models for acetylcholinesterase inhibitors based on KNN (1-nearest neighbors), and a quantitative prediction model based on support vector machine regression were used. The interaction of the compounds and receptors was analyzed using the molecular simulation method. Results: The radial basis kernel function was selected as the kernel function for support vector machine regression, and a total of 19 descriptors were selected to construct the quantitative prediction model.