COLLOQUIUM
Department of Computer Science and Engineering
University of South Carolina
 
Computational Methods in Structural Genomics:
High Throughput Protein Structure Determination from NMR Data

Homayoun Valafar

Departments of Computer Science and Biochemistry
University of Georgia

Date: March 26, 2004 (Friday)
Time: 3:30-4:30PM
Place: Swearingen 1A03 (Faculty Lounge)

Abstract

One of the primary aims of the structural genomics initiative is the
determination of representative structures from each protein fold
family. However, analysis of the current rate of discovery of new fold
families provides an estimate of 20-30 years of intense structure
determination before the completion of a full library. This is mainly
due to the inefficient nature of the target selection within the
structural genomics community. Consequently, it is important to
rapidly identify proteins that belong to a family that is already well
populated (so they can be eliminated from further studies), or more
importantly identify proteins that represent new families of
fold. Currently most prominent method of target selection is based on
sequence homology. While the utility of sequence homology in target
selection is the most efficient method of coverage of the sequence
space, it is not necessarily the most efficient method of covering
protein fold space.

During this presentation, the main objective of structural genomics
will be introduced. Discussion of some of the current existing
bottlenecks will follow the introduction. Finally, a new method for
rapid classification to a fold family via the utilization of signal
processing and optimization tools will be presented. Our statistical
analysis takes advantage of new source of data, namely residual
dipolar couplings that can be obtained by nuclear magnetic resonance
spectroscopy. The required NMR data can be quickly acquired and
analyzed. Using this method, structure determination efforts can be
focused on more unique and interesting structures, and the overall
efficiency in the construction of an information-rich structural
library can be increased.

Homayoun Valafar is the project coordinator for bioinformatics and
data analysis at the Southeastern Collaboratory for Structural
Genomics (SECGS) at the University of Georgia.