There are now over 13,000 Structural Genomics (SG) protein structures deposited in the Protein Data Bank (PDB). Most of these are of unknown or uncertain biochemical function and prediction of the function of proteins is an important current problem. Common approaches are informatics based. Our approach is based on functional residue prediction with Partial Order Optimum Likelihood (POOL), which is in turn based on computed chemical and geometric properties, as well as evolutionary history. The computed proton transfer behavior of the ionizable residues, obtained from the computed electrostatic potential function of the solvated protein molecule, is a key feature of the POOL method. Protein function is predicted using Structurally Aligned Local Sites of Activity (SALSAs). SALSA begins with consensus signatures, sets of predicted active residues for proteins of known, common biochemical function. Then the predicted sets of active residues for the SG proteins are compared and scored against the consensus signatures. A 3D structure match at the predicted, local functional site serves as a more reliable predictor of biochemical function than an overall structure match. The method is validated with sets of well-characterized proteins with similar folds but different functions. For select cases, predictions are validated experimentally by direct biochemical assay. Selected examples from structural genomics are shown, including instances where the previously reported annotation is incorrect. Examples where the SALSA prediction differs from that of the more conventional informatics-based prediction are highlighted. Supported by NSF-CHE-1305655.

Physicochemical properties for functional genomics: Predicting enzyme function for structural genomics proteins

251st National Meeting (2016)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.