| The data of protein sequence in the database is very large, and the protein structure prediction problem has become a very important issue in bioinformatics. It has a very important significance in the biology and medicine. In the prediction of protein folding problem, we first need to establish the mathematical model, and then use a better algorithm to simulate. Up to now, the protein Hydrophobic-Polar (HP) model is the most widely used and simplest model in simulation of protein folding problem.An existing algorithm is to add a curl operation (called Pull-Move) in the sparse area of the gird into the standard genetic algorithm for the protein folding of HP model. We define a Parallel-Vertical variation in the dense area of the grid. This variation makes up for the defect of Pull-Move which makes a mutation in the case of a vacancy, and the algorithm with this mutation is a global optimization algorithm. First, we tries to find the optimal energy of the eleven standard benchmark sequences by using the existing algorithm, and then we use our Parallel-Vertical variation to further optimize the results. The experimental results show that we can obtain better protein energy in the case of large length of protein. In particular, for the length of not less than sixty sequences, the results obtained by the genetic algorithm with pull-move are greatly improved, and with the increase in the length of the sequence, this improvement is obvious, and the number of changes show an increasing trend. In addition, for two sequences of length100, with the parallel-vertical variation, the change in energy even exceeds the optimal values obtained by genetic algorithm with pull-move and makes the optimal energy toward a theory of optimal value.The paper is organized as follows:the first chapter is the introduction, which briefly introduces the research content of bioinformatics and protein structure with its prediction methods. The second chapter introduces the Monte Carlo method and shows how to predict the structure of the protein2D HP model. The third chapter shows how the Genetic Algorithm predicts the structure of the protein2D HP model. The fourth chapter introduces our proposed algorithm and the numerical experiments with the new algorithm. The last chapter is the conclusion. |
