A heuristic method to bias protein's primary sequence in protein structure prediction

摘要

Protein Structure Prediction (PSP) is one of the most studied topics in the field of bioinformatics. Regarding the intrinsic hardness of the problem, during last decades several computational methods mainly based on artificial intelligence have been proposed to approach the problem. In this paper we broke the main process of PSP into two steps. The first step is making a bias in the sequence, i.e. providing a very fast yet considerably better energy of conformation compared to the primary sequence with zero energy. The second step, which is studied in the other essay, is feeding this biased sequence to another algorithm to find the best possible conformation. For the first step, we developed a new heuristic method to find a low-energy structure of a protein. The main concept of this method is based on rule extraction from previously determined conformations. We'll call this method Fast-Bias-Algorithm (FBA) mainly because it provides a modified structure with better energy from a primary (linear) structure of a protein in a remarkably short time, comparing to the time needed for the whole process. This method was implemented in Netlogo. We have tested this algorithm on several benchmark sequences ranging from 20 to 50-mers in two dimensional Hydrophobic Hydrophilic lattice models. Comparing with the result of the other algorithms, our method in less than 2% of their time reached up to 62% of the energy of their best conformation.