Extending Time Scales for Molecular Simulations

Abstract

Molecular dynamics simulations and protein structure prediction are essential tools in the quest for better understanding of the behavior of complex proteins. Based on these insights, new drugs can be designed to help fight diseases. Here we present a GPU implementation and analysis of adaptively restrained particle simulations (ARPS), which is up to fourteen times faster than conventional full dynamics molecular dynamics implementations on the GPU. Furthermore, we present a protein deformation algorithm which allows interactive manipulation of macromolecular graphs in accordance with popular force fields such as CHARMM, based on recent advances in graphics free form deformation. These two algorithms allow for longer and larger simulations, thus greatly enhancing the power of molecular simulations.