Impact of grid computing in structural biology
Ashley M. Buckle · Grid Computing · 2009
Structural biology aims to understand the function of proteins and other large molecules by determining their atomic structures in 3D. The technique of protein crystallography is typically used to solve this ~10,000 piece 3D jigsaw puzzle. One such crystallography method, termed Molecular Replacement (MR), uses known structures that are predicted to share some degree of structural likeness to the target, to kick-start the puzzle-solving process. This can be a trial-and-error procedure involving testing tens to thousands of starting structures in parallel, placing demands upon computational resources. In order to address this problem we have developed a hierarchical grid-based approach that leverages a range of distributed computational resources. Generally, the approach performs multiple MR calculations across a grid of networked computers, permitting high-throughput MR. We have leveraged three different classes of resource that were available to us, namely local research laboratory based computers; a 1000 CPU Condor pool at Monash University; and a World Wide Grid of machines leveraging computers in an Australian University Enterprise Grid, the PRAGMA testbed, the Open Science Grid and the Swiss National Grid. This has allowed us to perform high throughput MR calculations, thereby increasing the likelihood of determining the atomic structures of proteins.