Um algoritmo genético paralelo para o problema de dobramento de proteínas utilizando o modelo 3DHP com cadeia lateral

Abstract

This work presents a parallel genetic algorithm (PGA) for the protein folding problem, using the 3DHP-SC model. This model has been sparsely studied in the literature due to its complexity. A new fitness function was proposed, based on the free-energy and compacity of the folding. Special genetic operators were developed, besides strategies to aid the algorithm in the search of protein conformations. Many experiments were done to adjust all the parameters of the system, including the basic parameters of the GA (mutation and crossover probability, and tournament size) and parameters of the special genetic operators and strategies. The effect of the energy matrix of the model in the performance of the algorithm was also studied. Moreover, a comparison with other evolutionary computation approach was done, to verify the performance of the proposed method. Since there is no benchmark available to date, a set of 25 sequences was used, based on a simpler model. Results show that the PGA achieved a good level of efficiency and obtained biologically coherent results, suggesting its adequacy for the problem.