With advances in Structural Genomics and Proteomics, it becomes increasingly important to efficiently characterize the dynamics of proteins and their complexes because dynamics appears to be a natural link between structure and function. Conventional simulations are usually limited to small proteins and/or short times, due to computing limitations of time and memory, and sampling inefficiencies. A challenge is to develop both computationally efficient and physically realistic models and methods for estimating the collective dynamics of large structures and assemblies; and, in particular to assess the cooperative motions that are relevant to key biological functions. We have recently developed such a structure based computational approach to predict dynamics, referred to as the Gaussian Network Model (GNM). Within the scope of the present proposal, we will execute a computational methodology to overcome the limitations of GNM and enable us to rapidly predict the collective dynamics of proteins and larger complexes (Aim 1). We will validate the approach by comparison with experiments and simulations for both well-studied systems (eg. hemoglobin, Hk97 capsid, HIV-1 reverse transcriptase) and for a series of protein-inhibitor complexes, thereby increasing our understanding of the molecular machinery and key interactions underlying biological function (Aim 2). Finally, we will provide an on-line accessible, user-friendly computational interface to this methodology for biologists and biomedical scientists to explore structure -> dynamics -> function (SDF) relationships (Aim 3). This framework will include software using the protein structure database (PDB) as input, and providing as output predictions of biomolecular dynamics and key residues that should be targeted for effective control of dynamics. In addition to automated characterization and visualization of global dynamics, an interoperable, scalable database of results will be constructed, which will augment the information content of the rapidly accumulating structural data. This novel computational methodology will fill a unique niche due to its applicability to large structures and assemblies, its speed and accessibility to the scientific community.
Showing the most recent 10 out of 14 publications
