Case Study - Protein Folding In A Distributed Computing Environment

Project Leader: Marty Humphrey, University of Virginia;
Project Leader: Charles L. Brooks III, The Scripps Research Institute
Project Manager: Marty Humphrey, University of Virginia

Accurate modeling of protein folding is key to understanding the functions of cellular proteins.  The biomolecular simulation community primarily uses the molecular dynamics codes called CHARMM (Chemistry at Harvard Molecular Mechanics) and AMBER (Assisted Model Building with Energy Refinement) to explore protein folding.

CHARMM provides a general simulation environment for studies of the motions and mechanics of biomacromolecular systems.  It is supported across a variety of platforms, from Beowulf clusters to large parallel supercomputers such as the IBM Blue Horizon at the San Diego Supercomputer Center.  By grid-enabling these community codes, biologists will soon be able to conduct month-long simulations in only days from a Web-based interface.

A primary goal of this alpha project is to enable scientists to execute protein folding codes on a variety of compute engines by utilizing a portal "look-and-feel".  We are utilizing the Grid infrastructure across NPACI resources by which to execute CHARMM, Amber, NAMD, etc.  We want to effectively schedule these large parallel and sequential codes on the most appropriate resource at the time and to provide a robust, easy-to-use environment.

This project is currently using the following components from the NPACKage distribution:

• Globus
• Condor-G
• MyProxy
• SRB
• GridPort

We also have plans to use NWS and MPICH-G in the near future.  We will be running experiments on all NPACI Grid resources as well as our own cluster at the University of Virginia.