All-Atom Contact Analysis In Improving Structure Quality
Duke University, Durham NC
Investigators
Linked publications & trials
Abstract
The overall goal of this project is further development of all-atom contact analysis and its application to the improvement of protein structure determination, especially emphasizing automation needed for structural genomics. The project makes use of a recent methodological breakthrough by this laboratory which allows both visualization and quantitation of the steric details of molecular interaction, including all explicit H atoms. The visualization produces two patches of dot surface wherever two atoms are within 0.5Angstrom units of contact; the quantitative score has positive terms for H-bonds and van der Waals contact and negative terms for unfavorable overlap, evaluated locally on the surface. We have already produced an improved rotamer library and tools to display all- atom contacts along with electron density in crystallographic rebuilding programs, which can sensitively detect any fitting errors. The theoretical component of this project will work to reconcile the high level of packing specificity seen here with the predictions of standard energy-based models. The applied component will utilize the independent new information provided by this technique (not included in current refinement procedures) both to speed up and to improve the accuracy of structure determination by: automatically correcting amide flips and other systematic errors; incorporating improved loop libraries; automating sidechain fitting, including alternate conformations; developing better search strategies for optimizing sidechain and peptide flips, rotamer choices, and concerted motions during refinement; improving treatment of bond angle distortions around C-alphas; and developing convenient all-atom validation tools. Particular emphasis will be given to the needs of high-throughput crystallography, by smoothly integrating automated versions of these techniques into the systems used for the model-building, refinement, and validation stages of structural genomics work.
View original record on NIH RePORTER →