Genetic algorithm with advanced mechanisms applied to the protein structure prediction in a hydrophobic-polar model and cubic lattice

Abstract

This paper presents a genetic algorithm applied to the protein structure prediction in a hydrophobic-polar model on a cubic lattice. The proposed genetic algorithm is extended with crowding, clustering, repair, local search and opposition-based mechanisms. The crowding is responsible for maintaining the good solutions to the end of the evolutionary process while the clustering is used to divide a whole population into a number of subpopulations that can locate different good solutions. The repair mechanism transforms infeasible solutions to feasible solutions that do not occupy the lattice point for more than one monomer. In order to improve convergence speed the algorithm uses local search. This mechanism improves the quality of conformations with the local movement of one or two consecutive monomers through the entire conformation. The opposition-based mechanism is introduced to transform conformations to the opposite direction. In this way the algorithm easily improves good solutions on both sides of the sequence. The proposed algorithm was tested on a number of well-known hydrophobic-polar sequences. The obtained results show that the mechanisms employed improve the algorithm's performance and that our algorithm is superior to other state-of-the-art evolutionary and swarm algorithms.