Prediction of Protein Structure and Function Using Graphical Learning Biology Techniques. (360G-Wellcome-096622_Z_11_Z)

The accurate prediction of biological function on a genome-wide scale promiseswide-ranging benefits in understanding complex biological processes. Such understanding will be a key stepping stone in the development of techniques and pharmaceuticals to target genes associated with disease and their products. Historically, structural insights have provided the most detailed information on biological function, highlighting, for example a specific catalytic mechanism or interactions with partner proteins or other molecules. An in-depth analysis of key functional regions such as those in enzyme active sites, protein-protein binding sites, metal binding sites, and ligand binding clefts, as well as interacting regions between proteins is likely to add significantlyto the repertoire of tools currently available. In this project we will apply some new techniques we have developed to allow correlated mutations in protein families to be detected to the joint problem of protein structure and function By detecting pairs of sites which appear tochange in concert across the evolutionary tree we can infer the possible fold of the protein, residues involved in protein-protein interactions, and residues involved in ligand binding. From this information we hope to recognise a wide range of ligand binding sites, including DNA, RNA, A(TD)P, G(TD)P, sugars, cofactors and even other proteins. For the function prediction work, we may focus initially on predicting DNA and RNA binding sites which are of interest in Dr Savva's lab.