Computing statistical profiles of active sites in proteins

Active sites in proteins are three dimensional substructures that cause them to perform their function. The problem of finding substructures in a protein that are "similar" to the active sites of another protein has several important applications in biological sciences such as drug design, genetic engineering, arid diagriostic tools for arialysis of gerietically erigirieered pathogeris. Active sites cari be grouped irito families whose members are related by similarity of their furictioris. Iri this paper, we adapt Profile Hidderi Markov Models (PHMMs) to statistically profile active site families. We develop a serializatiori of the three dimerisiorial active sites that captures certairi shared physico-chemical and geometric features of the family. Experimerital results with our PHMM based method for profilirig active sites suggest that it is effective iri practice.