A novel approach for protein structure prediction based on an estimation of distribution algorithm

Abstract

Protein structure prediction is one of the major challenges in structural biology and has wide potential applications in biotechnology. However, the problem is faced with a difficult optimization requirement with particularly complex energy landscapes. The current article aims to present a novel approach namely AHEDA as an evolutionary-based solution to overcome the problem. AHEDA uses the hydrophobic-polar model to develop a robust and efficient evolutionary-based algorithm for protein structure prediction. The method utilizes an integrated estimation of distribution algorithm that attempts to optimize the search process and prevent the destruction of structural blocks. It also uses a stochastic local search to improve its accuracy. Based on a comprehensive comparison with other existing methods on 24 widely used benchmarks, AHEDA was shown to generate highly accurate predictions compared to the other similar methods.