The structure of protein assemblies is largely determined by non-covalent interactions between proteins. The importance of shape complementarity of interacting molecular surfaces is generally recognized, but is not easily studied, due to difficulties in quantifying shape. This proposal will develop methods for (1) characterizing the shape of protein surfaces, and (2) predicting complexes between proteins of known three- dimensional structure. The methods will be designed to work, even if the surface amino acid side chains have indeterminate orientation, as is often the case in protein crystallographic studies. Conformational flexibility will be modeled as partial occupancy (density) on a three- dimensional grid. The side chains' averaged density will be used for shape measurement. A reliable protein-docking computer program will be a quicker and less expensive means to study a protein association than an X-ray crystallographic experimental study of the complex. Applications include: (a) predicting how a monoclonal antibody binds to its protein antigen, (b) the assembly of viral coat protein monomers into capsids, and (c) the genetic disease sickle cell anemia, where hemoglobin monomers associate in an incorrect way. The software developed will be licensed to academic, governmental and industry scientists researching the molecular basis of disease.
