The combination of principal component analysis, genetic algorithm and tabu search in 3d molecular similarity

Abstract

Recently, molecular similarity, as an important tool of computer-aided drug design has developed rapidly. Its calculation has also developed from planar, rigid, 2D molecules to steric, flexible, 3D molecules. However, 3D molecular similarity calculation is easy to fall into local optima and the calculation is always time-consuming. In this paper, a method of flexible 3D molecular similarity calculation through the evaluation of molecular electrostatic potentials (MEP) with principal component analysis (PCA), genetic algorithm (GA) and Tabu search (TS) was presented. PCA was used to preprocess, GA was used to align two molecules and TS was used to decrease the probability of falling into local optima. The authors calculated the molecular similarities of benzene and its derivatives, a group of insecticides and a series of acetylcholinesterase inhibitors. To further evaluate the method, the authors calculated the similarities of HIV-1 protease inhibitors TIBO derivatives and predicted the pIC50 values, in which the linear relationship between similarities and logP values was also discussed.