Protein Structural Class Prediction Using Differential Evolution

Abstract

protein structural class prediction has been a challenging problem in protein science for many years. In this paper we present a new optimization approach using the Differential evolution (DE) for predicting the protein structural class. It uses the maximum component coefficient principle in association with the amino acid composition feature vector to efficiently classify the protein domains. The effectiveness is evaluated by comparing the results with that obtained from other existing methods using a standard database. Especially for all α and α +β class protein, the rate of accurate prediction by the proposed methods is much higher than their counterparts.