Although knowledge-based statistical potentials continue to be employed in successful protein structure prediction algorithms, two central components of knowledge-based approaches have been widely questioned.  1) Knowledge-based approaches only approximately address the correlations between interacting residues. 2) The physical significance of correlations within the PDB remains obscure because the PDB is not an equilibrium ensemble.  We have derived a framework that directly addresses these two shortcomings.  We have generalized the Yvon-Born-Green theory to exactly address the many-body correlations present in proteins.  Moreover, this theory is compatible with our extended ensemble framework for deriving transferable CG potentials.  In combination, these two theories provide a �first-principles� theory for determining  optimal transferable CG potentials directly from PDB structures. The fundamental assumption is that for each protein, the structures present in the PDB have been sampled according to their equilibrium distribution in solution.

Generalized Yvon-Born-Green theory for extended ensembles: 
A framework for calculating transferable potentials from known structures

239th National Meeting (2010)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.

Generalized Yvon-Born-Green theory for extended ensembles: A framework for calculating transferable potentials from known structures