ACADEMIA

TACC, IBM and World Community Grid celebrate research milestones

In late February 2010, World Community Grid — a network of volunteer computers comparable to one of the world’s top fastest supercomputers and dedicated to solving some of our most pressing scientific questions — reached the milestone of 500,000 registered users.

Harvesting the number-crunching power of off-duty computing cycles, World Community Grid helps researchers seek solutions to global problems, from the impact of climate change on Africa to cures for cancer, influenza, dengue fever, and hepatitis C. Together, registered users and the nearly 1.5 million connected devices, have accomplished more than 320,000 years of research in service to society. World Community Grid celebrated its 5th anniversary in November 2009.

The project reflects increasing efforts by academia and corporate America to channel the power of ubiquitous technology to solve Grand Challenge problems. As the corporate sponsor, IBM donated hardware, software, technical services and expertise to build and maintain World Community Grid.

“World Community Grid provides scientists with free supercomputing power that would otherwise not be available to them,” explained Sandy Dochen, IBM manager of corporate citizenship and corporate affairs. “If we combine technological innovation with visionary scientific research and large-scale volunteerism, we can all help make the planet smarter."

World Community Grid allows its distributed network of 1.5 million computers to work on scientific problems by dividing the workload of massive calculations into small pieces that can be calculated on a typical personal computer and then sent back to the Grid for reassembly. While there are other volunteer networks available, such as SETI@Home and Folding@Home, World Community Grid is the largest distributed volunteer network dedicated to humanitarian research.

World Community Grid has helped a dozen research projects to date (see box below or visit: www.worldcommunitygrid.org for a complete list) and has legitimized the use of volunteer computers for high-level science.

Since 2006, the Texas Advanced Computing Center (TACC) has supported the World Community Grid both as a resource provider — TACC’s Stampede system is the #14 contributor worldwide — and as an active collaborator on several of the research projects.

“TACC has been a magnificent partner,” Dochen said. “They get it and have been great about supporting World Community Grid.”

“Developing Dengue Drugs - Together” is one example of a World Community Grid project supported by TACC. The project seeks to find drug-like molecules that will be effective at stopping the replication of dengue, West Nile, and hepatitis C viruses, all of which share structurally-similar enzymes.

According to Stan Watowich, principal investigator and associate professor of Biochemistry at The University of Texas Medical Branch, the project represents one of the largest structure-based drug discovery projects ever attempted. TACC offers Watowich’s project volunteer cycles and dedicated time on its systems to set up the problem, to act as a data repository, and to perform pre- and post-processing on the data.

In Phase 1 of the research, three million compounds were tested using software based on AutoDock to determine which virtual molecules would bind to the surface or pockets of over a dozen virus proteins. World Community Grid computers returned several thousand potential matches.

“Phase 1 provided a crude ranking of small drug-like molecules,” said Watowich. “It allowed us to discard compounds that clearly aren’t going to fit into the active sites of the virus proteins.”

However, even among the compounds that do fit, most won’t bind to the virus proteins. From several dozen compounds tested by Watowich in the lab, only one or two had a significant inhibitory effect, meaning that 95 percent of the results from Phase 1 are likely “false positives,” according to Watowich.

Phase 2 will weed out the false positives by simulating the binding free energy of the well-fitting compounds using CHARMM. The process will differentiate the compounds that are likely to bind to the protein from those that would prefer to remain in the solvent. Watowich expects that the second round of World Community Grid simulations will return 100 candidate molecules with a high likelihood of being drug leads against dengue virus.

Large-scale, structure-based drug discovery — the Holy Grail for researchers and pharmaceutical companies alike — is rarely attempted because of the amount of computing power required to find probable compounds.

“These accurate free energy calculations are hardly ever done in drug discovery because they’re just so computationally expensive,” said Watowich. “But they’re currently the best way to predict if small molecules will bind to proteins.”

For Watowich, World Community Grid’s extreme scale allowed him to accomplish his computations at an accelerated pace and at no cost. In the past year, between 30,000 and 100,000 personal computers have worked on Watowich’s problem — the equivalent of having a supercomputer with 15,000 processing cores devoted exclusively to the dengue problem. (Note: The largest supercomputers currently have more than 100,000 cores.)

“If we’re able to significantly reduce the false positive rate and realize the hopes of computer-aided drug discovery, then we’ll have a real game-changer on our hands,” said Watowich.

Drugs derived from Watowich’s simulations could help millions worldwide who suffer from dengue, West Nile and hepatitis C, and have positive ramifications for other infectious diseases as well.

“We think this method can make a huge difference for world health,” said Watowich. By proving that crowd-sourced computer cycles can enable the discovery of viable drug compounds, the project sets the stage for further grid-based discoveries.

As World Community Grid continues to grow, the pace of the research supported by the Grid accelerates too. Phase 1 of the “Developing Dengue Drugs - Together” project took more than a year, but Watowich expects to deliver results from Phase 2 within six months.

Without the assistance of World Community Grid, such computations would have taken thousands of years.

“Imagine these researchers who often have to wait in line to access continuous computing power to crunch their data. With World Community Grid, this sometimes-trickle of information becomes a flood,” said Dochen. “It truly speeds up researchers’ ability to make progress for science.”