A Hybrid Population based ACO Algorithm for Protein Folding

A hybrid population based Ant Colony Optimization (ACO) algorithm PFold-P-ACO for protein folding in the HP model is proposed in this pa- per. This is the flrst population based ACO algorithm in the bioinformatics. It is shown experimentally that the algorithms achieves on nearly all test sequences at least comparable results to other state of the art algorithms. Compared to the state of the art ACO algorithm PFold-P-ACO slightly better results and is faster on long sequences. Proteins are one of the most important classes of bio- logical molecules. Chemically, a protein is a chain where each element is one of only 20 difierent amino acids. Each amino acid consists of a central carbon atom bonded to an amino group (NH2), a carboxyl group (COOH) and a side chain or residue (R) Hence, the amino acids dif- fer only in the residue R. One of the most important difierences between the residues is their hydrophobicity, i.e., how much they are repelled from a mass of water. The properties of the residues together with the environ- ment are responsible that the protein chain folds into a complex conformation. This conformation is called the "native" conformation of the molecule. The native con- formation is thermodynamically stable, i.e., it has small Gibbs free energy, and is very important for the function of the protein. The structure of a protein can be described on difierent levels: the amino acid sequence is the primary structure, the secondary structure describes characteristic struc- tures of the backbone of the molecule within local regions (e.g., alpha-helices or beta-sheets), the tertiary structure refers to the entire 3-dimensional structure. Difierent types of algorithms have been developed to predict the tertiary or secondary structure of proteins. All these algo- rithms use a model that is an abstraction of real proteins and describes important characteristics. An important class of models are the lattice models. A lattice model consists of a lattice that describes possible positions for the amino acids and an energy function that is to be min- imized and depends on the positions of the amino acids on the lattice. The most simplest lattice model is the HP model which is based on the observation that hy- drophobic forces are very important factors that drive the protein folding process. Advantages of the HP model are simplicity, that it shows several aspects of real pro- teins, and remains the hardness features of the biological problem. In this paper we propose a P-ACO algorithm called PFold-P-ACO for solving the protein folding in the HP model. PFold-P-ACO is the flrst P-ACO algorithm for the problem domain of bioinformatics. Section 2 describes the HP model and mentions some heuristics form the literature for the protein folding prob- lem in the HP model. An introduction to ACO and ACO approaches for the protein folding problem is given in Sec- tion 3. Population based ACO and our algorithm PFold- P-ACO are described in Section 4. The experiments and the results are presented in section 5. Conclusions are given in Section 6.