Protein 3D Structure Prediction By Improved Tabu Search

Since functions of proteins follow their three-dimensional structures to a large extent, understanding the protein structure information is significant for the study of protein function. Based on the minimum energy hypothesis that protein native structures are conformation at the global minima of their accessible free energies, some theoretical computing methods have been applied for protein structure prediction from its sequence of amino acids. At present, two important and difficult aspects are studied by many scholars. One is to design the energy function that can distinguish between protein natural and non-natural structure. The other is to find global optimization algorithm that can search the global minimum.Based on prominent simplified model—off-lattice AB model, the paper analyzes the advantages and disadvantages of Tabu Search algorithm. Then an improved Tabu search algorithm for protein three-dimensional structure and some improved strategies are proposed in this paper. For example, an effective heuristic strategy is proposed to generate initial conformations. Disturbance mutation strategy in Genetic Algorithm is applied to generating neighbor solution. Reasonable tabu conditions are designed. Improved Tabu search algorithm has been implemented by Java language and applied for 3D prediction of four Fibonacci sequences and three real protein sequences from Protein Data Bank (PDB). Experimental results show that the lowest energies computed by the improved TS algorithm are better than those obtained by the previous methods. At the same time, all the lowest-energy conformations form single hydrophobic cores in real proteins and hydrophobic monomers are always flanked by hydrophilic monomers along the chain which reflect the real protein reasonably. The improved TS algorithm has high accuracy and good convergence and can be effectively used to predict 3D structure prediction of proteins.