|
In our early studies, we have developed a computational strategy for de novo protein structure prediction based on the chain growth algorithm. In the past year, in collaboration with the Snoeyink group, we have expanded on the earlier algorithm. Specifically, O’Brien and Snoeyink have developed a significantly faster (at least, an order of magnitude) modification of the original approach which allows local tessellation of the growing protein chain as opposed to global tessellation as in the original publications. Several test protein structures ranging in size from ca. 5- to ca. 100 residues have been subjected to simulations, with the average cRMSD ranging between 4 to 7 Angstrom. We are currently working on the improvement in the prediction accuracy by starting the simulation not from the N-terminus but from residues identified as possible protein hydrophobic core. Our preliminary observations suggest that this approach leads to improved prediction accuracy, and we plan to expand on this approach in the Year 5 of the grant.