Petascale Simulations of Large Scale Biomolecular Assembly

Gregory Voth, University of Chicago

Usage Details

Gregory Voth, Glen Hocky, Morris Cohen, Zachary Jarin, Aleksander Durumeric, Tamara Bidone, Zhi Wang, Alexander Pak, Jesper Madsen, Dudu Tong, Alvin Yu, Paul Calio

The rapid growth of supercomputing power has enabled computer simulations to provide an increasingly useful complement to conventional experimental techniques. One important application of computer simulations is the study of biological systems at a molecular level of detail. Even modern supercomputers, however, struggle to simulate cell-scale processes at the molecular level.The Voth group have recently developed software for very large-scale "coarse-grained" (CG) simulations of biological processes, addressing several key technical barriers that can otherwise seriously hinder the efficient use of CG models. This software will be used to study a number of large-scale biomolecular systems of current interest in close collaboration with experimental colleagues.

The software consists of a highly scalable parallel molecular dynamics engine, using sparse data structures and space filling curves to provide highly efficient CG simulations of extremely large-scale molecular systems. In particular, the MD engine efficiently "load-balances" all required calculations across large supercomputing resources - even for extremely heterogeneous particle distributions, such as "implicit solvent" CG models. These characteristics will be of great value in the study of key aspects of the HIV-1 viral lifecycle, and the behavior of the large-scale branched actin assemblies which are important for cellular function. The continued development of the CG software, and the results of the large-scale CG simulations themselves, will provide valuable leads into both advancing the state-of-the-art in computer simulations and also helping to direct experiments to further elucidate processes of great biological importance.