Extraction of Feature Values using Protein Molecular Surface Data and Prediction of Significant Spot for Preferential Binding of Ligands

Abstract

Proteins often bind to other compounds at the concave site which is referred to as a pocket, and it is known that the binding site is locally similar with proteins binding to the same compound. In this paper, we propose a new method to predict the important structures when binding to the ligand in the pocket (described as "significant spot"), by focusing on the distance between the surfaces of proteins. Experiments were conducted with 60 kinds of proteins to extract significant spot for each protein. As a result, significant spot were extracted with high accuracy, and the effectiveness of the proposed method was confirmed.