Copernicus: a new paradigm for parallel adaptive molecular dynamics
Biomolecular simulation is a core application on supercomputers, but it is exceptionally difficult to achieve the strong scaling necessary to reach biologically relevant timescales. Here, we present a new paradigm for parallel adaptive molecular dynamics and a publicly available implementation: Copernicus. This framework combines performance-leading molecular dynamics parallelized on three levels (SIMD, threads, and message-passing) with kinetic clustering, statistical model building and real-time result monitoring. Copernicus enables execution as single parallel jobs with automatic resource allocation. Even for a small protein such as villin (9,864 atoms), Copernicus exhibits near-linear strong scaling from 1 to 5,376 AMD cores. Starting from extended chains we observe structures 0.6 Å from the native state within 30h, and achieve sufficient sampling to predict the native state without a priori knowledge after 80–90h. To match Copernicus’ efficiency, a classical simulation would have to exceed 50 microseconds per day, currently infeasible even with custom hardware designed for simulations.
CONCLUSIONS
The parallel adaptive approach to molecular dynamics presented here is a powerful combination of the strongest aspects of parallel simulation, distributed computing, and Markov State Models. We have shown that the Copernicus framework effectively increases the scale at which biomolecular molecular dynamics simulations can be performed from a few hundred cores to many thousands and beyond, likely reaching millions of cores for large molecules. In contrast to custom hardware that parallelizes a single simulation, Copernicus achieves this by using domain-specific knowledge to focus on the desired results of biomolecular simulations: quantitatively robust estimation of molecular reaction parameters. Our approach uses statistical mechanics knowledge to drive a message-passing parallelization scheme, and adapts this scheme to match the available computational resources by maximizing CPU and bandwidth use simultaneously.
Copernicus provides an open { but authenticated { peer-to-peer architecture for ensemble simulations in high performance computing. Currently, Copernicus comes with plugins to run Markov-State-Model-driven sampling and Bennett Acceptance Ratio free energy perturbation calculations. In addition to the architectural improvements outlined above, further development will focus on making available controllers optimized for different sampling algorithms and questions in statistical mechanics modeling of biomolecules.
IMPLEMENTATION
Download at http://copernicus-computing.org/
Public git access
You can track development versions of Copernicus with the following git repository:
git clone git://git.copernicus-computing.org/cpc copernicus
Authors plan to keep the latest stable version in the master branch.
Sander Pronk, Per Larsson, Iman Pouya, Gregory R. Bowman, Imran S. Haque, Kyle Beauchamp, Berk Hess, Vijay S. Pande, Peter M. Kasson, and Erik Lindahl. 2011. Copernicus: a new paradigm for parallel adaptive molecular dynamics. In Proceedings of 2011 International Conference for High Performance Computing (SC11), Networking, Storage and Analysis (SC ’11). ACM, New York, NY, USA, , Article 60 , 10 pages. [DOI: 10.1145/2063384.2063465]
Category: Articles, Life Science, SC11, Software
Subscribe
If you enjoyed this article, subscribe to receive more just like it.
