Genome-wide protein structure prediction and structure-based function annotation
Thursday, March 8, 3:30PM – 5PM
The biological function of protein molecules is determined by the shape of their three-dimensional structures. Is it possible to predict protein structure and function from the amino acid sequence? We developed a new algorithm, I-TASSER, which assembles atomic structure of proteins using fragments excised from unrelated experiment structures. Functional insights (e.g. ligand-binding affinity, enzyme classification and gene ontology) are then deduced by matching the predicted structure models with the known proteins in protein function libraries. The I-TASSER algorithm was ranked as the best for automated protein structure prediction in the communitywide CASP experiments of 2006, 2008 and 2010; it was also ranked at the top for protein function annotation in CASP9 in 2010.
In this talk, we first review the recent progress in computer-based protein structure prediction including the new developments in ab initio folding and atomic structure refinements since the CASP9 experiment, and show that the protein structure prediction problem can in principle be solved using the current PDB library. Next, we discuss the application of the developed methods to the structural and functional modeling of a number of genomes, including all G-protein coupled receptors (GPCRs) in the human genome, yielding models 90% of which are shown to have correct topology, and Marek’s disease virus, the first success of the computational modeling of a complete viral genome. Finally, we demonstrate how the predicted I-TASSER structure models can be used to annotate the biological function of the proteins and screening drug candidates by matching their global topology and functional sites against the existing structure/function/binding databases.
Hosted by Ron Elber