Computational Biology and Bioinformatics Discussion Group
Wednesday, January 12, 2005
Presented by the Computational Bio & Bioinformatics Discussion Group
Organizer: Andras Fiser, Albert Einstein College of Medicine
This group brings together diverse institutions and communities to share new and relevant information at the frontiers of the interrelated fields of bioinformatics and computational biology.
Andrew Neuwald, Cold Spring Harbor Laboratory
"Measuring Evolutionary Constraints Reflecting Underlying Mechanisms of Ras, Rab, Rho and Ran GTPases."
Roland Dunbrack, Fox Chase Cancer Center
"Benchmarking and Improving the Accuracy of Comparative Modeling of Protein Structures."
"Measuring Evolutionary Constraints Reflecting Underlying Mechanisms of Ras, Rab, Rho and Ran GTPases"
Conserved patterns in multiple sequence and structural alignments are due to selective constraints imposed during protein evolution. These patterns correspond to functionally critical atomic interactions and thus contain implicit information about underlying protein mechanisms, just as crystalline X-ray diffraction patterns contain implicit information about a protein's structure. Contrast hierarchical alignment and interaction network (CHAIN) analysis uses rigorous statistical procedures to measure and characterize these constraints and attempts to fit mechanistic models to these constraints in a manner analogous to fitting an atomic model to the electron density inferred from X-ray diffraction data. This is illustrated through an analysis of P loop GTPases, which reveals that Rab, Rho, Ras, and Ran share a conserved network of molecular interactions centered on bound nucleotide. This network presumably performs a crucial structural and/or mechanistic role considering that it has persisted for more than a billion years after the divergence of these families. An analysis of the Ran family in the light of Ran co-crystal structures suggests detailed aspects of Ran's C-terminal, basic patch and nucleotide exchange mechanisms.
"Benchmarking and Improving the Accuracy of Comparative Modeling of Protein Structures"
Accuracy of comparative modeling methods is crucial if these methods are to be useful in understanding biological function and improving our ability to affect function through chemical methods. We have used large benchmarks to test methods in sequence alignment and database search, side-chain prediction, and loop prediction. In sequence alignment, we have recently optimized profile-profile alignment methods that include secondary structure information. Evaluating side-chain and loop prediction depends on the accuracy of experimental structures. Electron density calculations for a large set of proteins demonstrates that side-chain prediction improves remarkably for side-chains with higher electron density and less apparent disorder. We have identified a much larger set of side chains with multiple rotamers with clear occupancy than is identified in the PDB. These calculations can be used to improve rotamer libraries, and to provide test sets for evaluating prediction methods that include predicting side-chain disorder.