Physicochemical Methods for Prediction of Functional Information for Proteins

Structural genomics initiatives are determining thousands of new protein structures. Many of these structures are of unknown function, and computational methods for the rapid determination of functional information from protein structure are needed. We present details of how functional information is obtained from the structure using THEMATICS (Theoretical Microscopic Titration Curves). THEMATICS is a computational procedure that gives information about chemical reactivity, based on solution of the Poisson-Boltzmann equations for the electrical potential function. We show how anomalies in predicted titration curves are established. We show further that when residues with anomalous predicted titration curves form a cluster in physical space, these residues tend to be very highly conserved across species and such clusters are reliable predictors of the active site. Results are given for ten enzymes; detailed results are shown for the enzymes triosephosphate isomerase (from chicken), 6-hydroxymethyl-7,8-dihydropterin pyrophosphokinase (from E. coli), and papain (from papaya).