Theory and application of restricted five neighborhood cellular automata (R5NCA) for protein structure prediction

Abstract

This paper reports the theory of a special class of Cellular Automata (CA) referred to as Restricted 5 Neighborhood CA (R5NCA). Its application deals with identification of Protein Structure. Each amino acid of a protein chain is modeled with a R5NCA rule. In the process, a protein gets modeled as a R5NCA. CA Evolution models the evolution of the protein to its minimum energy folded configuration. The physical domain parameters are next mapped to CA model parameters from the analysis of known structural data available in Protein Data Base (PDB). The process of reverse mapping is implemented to identify the structure of a protein in blind test from its model parameters. The CA model achieves close to 99% correct prediction for secondary structure prediction, while for tertiary structure prediction, the average RMSD (Root Mean Square Deviation) value has been found to 1.82. © 2012 Springer-Verlag Berlin Heidelberg.