An Algorithm for Finding Similarities in Protein Active Sites

Abstract

A new method has been developed to find similar substructures in protein binding cavities. It is based on the idea that protein function is intimately related to the recognition and subsequent response to the binding of an endogenous ligand in a well-characterized binding pocket. It can therefore be assumed that proteins having similar binding cavities also bind similar ligands and exhibit related function. For the comparison of the binding cavities, the binding-site exposed physicochemical characteristics are described by assigning generic pseudocenters to the functional groups of the amino acids flanking a particular active site. These pseudocenters are assembled into small substructures. To find substructures with spatial similarity and appropriate chemical properties, an emergent self-organizing map is used for clustering. Two substructures which are found to be similar form the basis for an expanded comparison of the complete cavities. Preliminary results with four pairs of binding cavities show that similarities are detected correctly and motivate further studies.