New Strategies for Protein Folding
Joseph F. Danzer, Derek A. Debe, Matt J. Carlson, and William A. Goddard III
The RGP (Restricted Generic Protein) direct Monte Carlo algorithm has proven to be very efficient at generating the ensemble of protein topomers consistent with a set of inter-residue restraints for a given protein. Through a series of energy minimizations, to total ensemble of several thousand topomers can be quickly reduced to a set of ~50 of which one is certain to be the proteinís native topology. Investigations into using loose restraints (~10Å) based on the PHD secondary structure prediction algorithm are being done. Preliminary results for two proteins suggest that growing an ensemble of 5,000 structure with 3-4 loose restraints, is sufficient to be certain of obtaining the native topology in the final set of ~50 proteins. The RGP algorithm has proven to generate structures with CRMS of ~6Å from the native structure. Several tools are currently being developed in order to refine the structures obtained from RGP. These include a threading algorithm which will locally modify segments of the protein that are found to be homologous with other segments in the protein data bank whose structure is known, but ensuring that the overall topology is not disrupted. A dynamic Mote Carlo algorithm for generating local deformations in the protein backbone in order to maximize the number of hydrophobic side chains pointing inwards to the center of mass and the number of hydrophilic side chains point outwards from the center of mass..
This research has been funded by grants from NIH and NSF.
Figure 1. Generate and select hierarchy used to predict protein structure.
Figure 2. LexA repressor protein, predicted structure (dark) and the native structure (light).
Figure 3. Myoglobin, predicted structure (dark) and the native structure (light).
Author: Joseph Danzer