Geometric 3D protein structures superimposition

摘要

The present paper describes a 3D protein structure superimposition technique. The structures are considered as a cloud of points in 3D space, and the problem of comparing two protein structures can be reformulated as finding the largest common point set LCP. Many used methods are based on distances between C-alpha atoms to reach an alignment or by generating all possible local alignments to achieve a global alignment. Here we will present and discuss an implementation of the Discrete to the Continuous approach. This one performs the superimposition task without any initial alignment (initial correspondence is not required). The experimental result was accomplished using 460 tests and gives rise of promising performance.