Manifoldized beta-shape을 이용한 단백질의 pocket 계산

Abstract

The study on the interactions between a protein and a compound could be facilitated based on the identified pockets on a protein and the pockets are important for many applications. Hence, identification of pockets on a protein is an important issue in the molecular biology. Pockets are usually recognized as some depressed regions on the boundary of proteins. Recently, computational approaches using geometrically well-defined structures have emerged for the pocket extraction. In this paper, we propose an algorithm for extracting the pocket on a protein using β-shapes. The β- shape is a geometric structure which contains the proximity among atoms on the boundary of proteins. Hence, we formulate the pocket extraction problem as the segmentation of meshes on the β-shape boundary. However, a β-shape is in general non-manifold and therefore it is not convenient to traverse topology, not to speak of the heavy weight data structure. The proposed algorithm first converts the boundary of β-shape into a manifold, called manifoldized β-shapes, by topological operators. Then, we use some topological properties of the entities on the β-shape boundary to segment it into a set of pocket candidates.