NEW YORK, Aug. 30 - A research team at the San Diego Supercomputer Center is using a grid-based computer system from Entropia to build a set of protein structure databases.
The SDSC team, led by Kim Baldridge, is building three databases, said Entropia director of applications Stephen Elbert. These include a small molecule set, a database of intermediate-sized protein-bound ligands, and a collection of component ligands relevant to potential bioterrorism agents.
Entropia's DCGrid harnesses the unused power of desktop computers to perform complex calculations. At SDSC, the desktop grid runs GAMESS software, which performs structural calculations and optimizes geometrical information.
With the help of 250 desktop computers with processing power ranging from 180 MHz to 2.2 GHz, the platform has so far completed calculations on almost 1,000 proteins, said Elbert.
"What they've done in principle they could have done on one of their supercomputers, but those machines are heavily used for other projects," he said. "This is a way of expanding capacity. And it's a whole lot cheaper."
Elbert said that the SDSC researchers hope to finish calculations on all the small molecules in the Protein Data Bank within the next few months.
While a PC-based grid system might not be able to keep pace with a high-end machine, grids grow more powerful as the average power of a desktop machine increases, said Elbert. "That's the beauty of this system: Desktop systems regularly get replaced."
Entropia, a San Diego-based computing company specializing in grid systems, anticipates that these networks will be particularly useful in chemistry and life sciences.
Compound screening are "a natural" for grid-based platforms, said Elbert. The company plans to make several new life sciences customer announcements in September.