Hill-climbing Search in Evolutionary Models for Protein Folding Simulations

Evolutionary algorithms and hill-climbing search models are investigated to address the protein structure prediction problem. This is a well-known NP-hard problem representing one of the most important and challenging problems in computational biology. The pull move operation is engaged as the main local search operator in several approaches to protein structure prediction. The considered approaches are: (i) a steepest-ascent hill-climbing search guided by pull move transformations, (ii) an evolutionary model with problem-specific crossover and pull move mutations, and (iii) an evolutionary algorithm based on hill-climbing search operators. Numerical experiments emphasize the advantages of the latter approach for several difficult protein benchmarks.