Enhanced Genetic Algorithm for Protein Structure Prediction based on the HP Model

Abstract

Proteins are organic compounds that are made up of combinations of amino acids and are of different types and roles in living organisms. Initially a protein is a linear chain of amino acids, ranging from a few tens up to thousands of amino acids. Proteins fold, under the influence of several chemical and physical factors, into their 3-dimensional structures which determine their biological functions and properties. Misfolding occurs when the protein folds into a 3D structure that does not represent its correct native structure, which can lead to many diseases such as Alzheimer, several types of cancer, etc... (Prusiner, 1998). Hence, predicting the native structure of a protein from its primary sequence is an important and challenging task especially that this protein structure prediction (PSP) problem is computationally intractable. The primary structure of a protein is a linear sequence of amino acids connected together via peptide bonds. Proteins fold due to hydrophobic effect, Vander Waals interactions, electrostatic forces, and Hydrogen bonding (Setubal & Meidanis, 1997). The secondary structures are three-dimensional structures characterized by repeating bonding patterns of ┙-helices and ┚-strands. Proteins further fold into a tertiary structure forming a bundle of secondary structures and loops. Furthermore, the aggregation of tertiary structure regions of separate protein sequences leads to quaternary structures. These structures are depicted in Fig. 1 (Rylance, 2004). Computational approaches for PSP can be classified as: homology modeling, threading, and ab initio methods (Floudas, 2007). Approaches in the first two groups use known protein structures from protein data banks (PDB). Approaches in the third group solely rely on the given amino acid sequence. A survey of PSP approaches appeared in Sikder and Zomaya (2005). Homology modeling uses sequences of known structures in the PDB to align with the target protein’s sequence for which the 3D structure is to be predicted (Kopp & Schwede, 2004; Notredame, 2002; Pandit et al., 2006). Threading is similar to homology modeling. But, instead of finding similar sequences to deduce the native conformation of the target protein, threading assumes that the target structure is similar to another existing structure, which should be searched for (Lathrop et al., 1998; Jones 1998; Skolnick et al., 2004). The threading of a sequence to a fold is evaluated by either environment-based or knowledge-based mean-force-potentials derived from the PDB.