Protein Tertiary Structure Prediction Using Evolutionary Algorithms

In this paper, Protein Tertiary Structure Prediction using Evolutionary Algorithms (EAs) such as SelfAdaptive Differential Evolution (SaDE) and Real-coded Genetic Algorithm (RGA) are discussed. RGA is implemented with various crossover and mutation operators. The algorithms are tested on a peptide Metenkephalin. The energy functions used are ECEPP/2 and ECEPP/3 force fields. SaDE and RGA with discrete crossover and boundary mutation produce the best energy values than other crossover and mutation operators of RGA. But, Statistical results of SaDE and RGA show that SaDE outperforms RGA in terms of number of function evaluations, mean energy and success rate. The best results obtained using SaDE and RGA are compared with native structure 1PLW and classical benchmark Scheraga conformation and the corresponding minimum RMSD values are 2.13 A o and 1.45 A o respectively. Comparison of the best results of SaDE and RGA with other reported RGA variants show better performance in terms of energy and computational search efficiency. A set of unique hundred best solutions obtained from both algorithms are clustered using hierarchical cluster algorithm. This gives seven independent clusters suggesting the robustness of these methodologies and the ability to explore the conformational space available and to populate the near native conformations.