Qualitative and Quantitative Aspects of Biomolecular Systems Revealed by Large Scale and Grid Computing Methods

Abstract

Biomolecules, for example proteins, are dynamic entities and their interactions are important in determining the behaviour of cells. We shall use computational methods, specifically classical molecular dynamics, to simulate the dynamics of several biomolecular systems. The free energy of binding is the key quantitative measure of how strongly two molecules interact. Although theoretical methods exist for calculating these free energies, their use and validation has been hampered by the length of time it takes to compute a single value. We shall describe a novel method, steered thermodynamic integration using molecular dynamics (or STIMD), that uses a computational grid to accelerate from months to less than one week the time taken to calculate a single difference in binding free energy (!!G) and its use to study the binding of a peptide to a v-Src SH2 protein domain. Two related mutations are studied and the calculated and experimental values of !!G are compared. We shall then discuss the insight into the binding event gained through the decomposition of the free energy by components of the system. The problems we encountered when using STIMD are described and several future developments are discussed. We shall discuss our protocol for integrating a prostaglandin H2 synthase (PGHS) monomer protein into a phospholipid bilayer and present evidence that the protein is correctly integrated. This is the first step in a qualitative study of the dynamical differences between the two isozymes of PGHS, the target for non-steroidal anti-inflammatory drugs, and we shall also describe several observed dynamical differences. We have used high-performance computing and computational grids for all these quantitative and qualitative studies as using classical molecular dynamics to simulate these biomolecular systems is extremely computationally intensive.