This award facilitates scientific research using the large, new, computational resource named Blue Waters being developed by IBM and scheduled to be deployed at the University of Illinois. It provides travel funds to support technical coordination between the principal investigators, the Blue Waters project team and the vendor technical team
The project implements an operational petascale multiscale methodology that can describe large length-scale biomolecular systems. The project specifically targets the following simulations of significant experimental interest: protein induced membrane remodeling, key steps of the HIV viral replication cycle and clathrin coated pit formation in endocytosis. A coupling of the petascale multiscale simulation with experimental input will serve to benchmark the quantitative predictive power of the simulation methodology while also providing new insight into experimentally unresolved behavior in these important biomolecular relevant systems.
The results of this research will lead to advances in biomedical research related to disease such as HIV infection and replication, thereby directly affecting human health and welfare. The project team is led by researchers from the fields of theoretical/computational chemistry and biophysics and the team also includes graduate students and undergraduates. It offers an opportunity for students to play a role in establishing the U.S. as the leader in petascale biomolecular multiscale simulation.
Computational molecular simulations have become an important scientific tool, providing detailed information that is otherwise difficult or impossible to examine with conventional experiments. Even with modern supercomputers, however, molecular dynamics (MD) simulations that may involve billions of interacting particles over many millions of small "time steps" can be exceedingly difficult to perform. As many phenomena of scientific interest occur in these relatively large and complicated molecular systems, significant barriers remain to the application of computer simulations in many key areas of the physical, medical and materials sciences. One promising approach to examine very large molecular systems is "coarse graining" (CG), a process which generates simplified, computationally efficient representations that nontheless capture fundamental molecular properties. CG models can greatly extend both the length and time scales accessible to computer simulations, particularly where the effects of solvent molecules (such as water) are incorporated directly into the properties of the other molecules (an "implicit solvent" model). The use of CG models with implicit solvent is not yet a universal panacea, as the nature of the CG models themselves can introduce particular challenges to the efficient use of supercomputers. The Voth research group, long at the forefront of CG molecular simulations, recently developed the theory of "ultra-coarse-grained" models (UCG). UCG techniques enable not only the simulation of the usual molecular motions in space, but also capture the atomic-resolution behaviors that can dynamically change CG molecule properties during a simulation. In order to apply UCG techniques to large-scale molecular phenomena, the Voth group designed and implemented UCG-MD: an unorthodox, highly parallel molecular simulation engine designed from first principles to address key weaknesses in existing software for CG simulations on modern supercomputers. The UCG-MD software algorithms were demonstrated to scale efficiently to very large numbers of CPU cores, even for traditionally problematic cases, with detailed descriptions of the key algorithms then disseminated to the wider scientific community. The UCG-MD code has thus provided new possibilities for computer simulation in advancing the the physical, medical and materials sciences. The UCG-MD code is being used to investigate a variety of phenomena of scientific interest, such as the important stages in the maturation process of the human immunodeficiency virus (HIV-1).
