Molecular flexibility is an intrinsic property of bio-molecules and is often essential for their function. Its understanding is important for relating structure to biological function, which is at the heart of computer aided drug design. The lack of high-level representation of molecular flexibility prevents current simulation methods from being able to predict complexes presenting drug-induced conformational changes in the receptor. This project proposes the development of a novel multiresolution encoding of molecular flexibility, based on a hierarchical, tree-like structure. Each node in this Flexibility Tree (FT) will represent a subset of atoms that move as a rigid body at a given level of approximation. Every node will store information about the motion of its children. Shapes assigned to child nodes will be used to define recursively the shape of parent nodes. Operators for the convolution of shape and motion, and for the evaluation of the fitness of conformations described by this tree will be developed. The docking program AutoDock will be extended to represent protein flexibility using the FT. This data structure will enable: 1) the high-level representation in computational models of macro- molecular flexibility as a combination of nested motions occurring at different scales, 2) the selective encoding of a conformational sub-space of interest, and 3) the study of the effects of flexibility on molecular shape and properties. The proposed research defines a novel framework for structural molecular biologists to analyze molecular structures and their interactions using a unified, multi-scale representation of both shape and flexibility information. It will help increase our understanding of the fundamental nature of biomolecular flexibility and provide the infrastructure for integrating protein flexibility in computational methods such as automated docking. It will also provide computational tools to explore, either interactively or programmatically, molecular flexibility at various levels of detail. These tools will be made available to the scientific community on a large number of computational platforms to ensure their wide dissemination. These tools hold the promise of extending dramatically the range of biological systems that can be simulated successfully, and thereby improve our ability to discover and optimize new drugs.
