HIERARCHICAL METH FOR GLOBAL MINIMIZ OF MACROMOL ENERGY: STRUC BASED DRUG DESIGN

Cornell University Ithaca, Ithaca NY

Investigators

Linked publications & trials

Abstract

Structure-based design of protein inhibitors requires the ability to predict conformational and free-energy changes occurring during protein-ligand binding. In principle this can be done by finding the conformation of lowest energy (global minimum) and the free-energy of the associated macrostate. But, in practice, the numerical complexity of the computation exceeds capabilities by many orders-of-magnitude. We are developing a new class of algorithms that uncovers and exploits the intrinsic hierarchical structure of macromolecular potential energy landscapes to thereby greatly increase the efficiency of this process. The approach, adapted from renormalization group methods which have been very successful in statistical physics, replaces conventional fixed-scale minimization and sampling with a sequence of operations at a series of decreasing size scales. During the past year we have developed much of the needed mathematical formalism and have performed tests on model systems including small peptides. We are currently extending our tests to larger peptides and developing the tools needed for application to larger proteins.