The main goal of this project is to obtain a method which will produce an accurate model of a protein-protein complex from two separate, flexible, and possible inaccurate protein structures. A distinguishing feature of this new docking procedure is that it searches through the conformation of each protein in addition to the relative positions of the two proteins. This project involves the further development of GRAMM (Global RAnge Molecular Matching) docking methodology and software for protein docking. GRAMM is used at low resolution to predict the structures of complexes in the non-redundant database of protein complexes using both actual individual structures from the complexes and their models. The flexible docking procedure is expected to be more accurate and more robust with respect to the inaccuracies in the initial approximate structures of the interacting proteins than are the currently prevalent rigid docking procedures. The approach will be especially useful in view of the rate of advance in the human genome project, since it will allow a rapid evaluation of functions for a large number of predicted protein structures. Thus, the combination of protein structure prediction and protein docking is expected to increase greatly the usefulness of molecular modeling.
