Pose Estimation of Inhibitor Binding by Point Cloud Registration and Molecular Surface Representation Optimization

Abstract

Proteins, which are enzymes, act only on specific substrates (chemical substances) to cause specific chemical reactions. Some of these reactions are harmful to the human body, which can be prevented by pre-binding the protein to another substrate (called an inhibitor). In this study, we propose a method to estimate the pose of the inhibitor when it binds to the pocket, the active site of the protein that causes a specific chemical reaction. The conventional method of point cloud registration using FPFH local features has low registration accuracy, and the accuracy of inhibitor search is insufficient. To solve this problem and improve the accuracy, we propose a method to solve the registration using global features and optimize the molecular surface representation, then confirm the effectiveness of the proposed method.