A New Method for Unbiased Comparison of Protein Structures

Abstract

While there exist several excellent algorithms for protein sequence comparison, the comparison of protein structures is more difficult This paper presents a method for a sequence-independent, unbiased comparison of protein structures. The method treats each protein as a collection of a carbons without any regard to sequence or chain-connectivity. Heuristics are used to select a series of subsets of atoms in each of the protein structures. These subsets of atoms are then compared using a subgraph-isomorphism search technique. Partial matches thus obtained are then used to superpose one protein onto the other. A simple scoring function is then used to determine the validity of the match This method has been applied to a set of about 100 diverse proteins. Several proteins known to have similar structures were identified by the method A few protein-pairs not widely recognized to have structural similarities were also identified.