An improved artificial bee colony algorithm for 3D protein structure prediction

Abstract

One of the key problems in the field of bioinformatics is protein structure prediction. The thermodynamic hypothesis demonstrates that protein's energy is the lowest in nature state. So protein's structure can be gotten directly by protein sequence's free-energy. In this paper, an improved algorithm based on three-dimensional AB off-lattice model to improve local search and global optimization ability of artificial bee colony algorithm has been presented. The simulation experiment shows that it can effectively search the lowest free-energy in the condition of keeping high accuracy. The experimental results indicate that the minimum energy from the improved artificial bee colony algorithm is better than other similar algorithms, and with the increase of protein sequence's length, this algorithm has better performance.