Compositional preferences in quadruplets of nearest neighbor residues in protein structures: statistical geometry analysis

Abstract

The 3D structure and amino acid sequence of proteins are related by an unknown set of rules that is often referred to as the folding code. This code is believed to be significantly influenced by nonlocal interactions between the residues. A quantitative description of nonlocal contacts requires the identification of neighboring residues. We applied statistical geometry approach to analyze the patterns of spatial proximity of residues in known protein structures. Structures from a dataset of well resolved nonhomologous proteins with a single point representation of residues by C/sub /spl alpha// atoms were tessellated using Delaunay algorithm. Compositional analysis of Delaunay simplices reveals highly nonrandom clustering of amino acid residues in protein structures. Results of the analysis are implemented in algorithms for protein structure classification and prediction.