Amino acid distribution rules predict protein fold

Abstract

In the present article, we provide a brief overview of the main approaches to analysing the sequence–structure relationship of proteins and outline a novel method of structure prediction. The proposed method involves finding a set of rules that describes a correlation between the distribution of residues in a sequence and the essential structural characteristics of a protein structure. The residue distribution rules specify the ‘favourable’ residues that are required in certain positions of a polypeptide chain in order for it to assume a particular protein fold, and the ‘unfavourable’ residues incompatible with the given fold. Identification of amino acid distribution rules derives from examination of inter-residue contacts. We describe residue distribution rules for a large group of β-sandwich-like proteins characterized by a specific arrangement of strands in their two β-sheets. It was shown that this method has very high accuracy (approximately 85%). The advantage of the residue rule approach is that it makes possible prediction of protein folding even in polypeptide chains that have very low global sequence similarities, as low as 18%. Another potential benefit is that a better understanding of which residues play essential roles in a given protein fold may facilitate rational protein engineering design.