EVENTS | VIEW CALENDAR
Supercomputing, one PC at a time
LA JOLLA, Calif.—Thanks in large part to IBM and hundreds of thousands of unassuming computer users, a team of scientists at Scripps Research Institute has trail-blazed some serious new ground in the fight against HIV infection and AIDS with a huge fragment-based screening experiment that has revealed two new binding sites for small molecules to potentially affect HIV protease.
This is important, notes team member Dr. Arthur J. Olson, a professor in the Scripps Department of Molecular Biology, because all current FDA-approved HIV protease inhibitors target the active site—the catalytic center of HIV, if you will. But this research, he says, points to the value of attacking HIV from an "exo-site" farther removed from the center of the action.
"If you can find secondary sites on the proteins that can be attacked and can change the behavior of the virus, you can develop entirely new drugs that can be used with existing ones to box HIV protease in," Olson says. "This could give you the ability to fight the virus even as it tries to mutate at the active site to counteract the inhibitors currently in use or in development."
The Scripps team's research is published in the March cover story, "Fragment-Based Screen Against HIV Protease" of the journal Chemical Biology and Drug Design.
The team's research in another area, producing a more accurate model of HIV integrase in order to improve research toward more clinically effective integrase inhibitors, is being published in the article "A Dynamic Model of HIV Integrase Inhibition and Drug
Resistance" in the March issue of the Journal of Molecular Biology.
Although protease and integrase are two different HIV enzymes, the Scripps work in these two areas shares a common denominator, and that's where IBM comes in—in the form of the IBM World Community Grid. Olson says that the Scripps research on both HIV fronts might not have even been possible without that technology—technology that relies on average personal computer users coming together to create a virtual supercomputer.
The concept is basically an amped-up form of grid computing, something that the Scripps team was already doing with its FightAIDS@Home project. But until IBM came along, the Scripps team struggled to build up a collection of volunteers that would give its grid computing system the oomph it needed to give Scripps researchers the computing power they needed.
Over the past five years, however, IBM's World Community Grid and its volunteers have allowed Scripps to run more than 100 million computations to evaluate the potential interactions between compounds and mutant viral proteins. Over that time, the World Community Grid has essentially provided Scripps researchers with 107,000 years of computation time, and helped shape theories on drug resistance and molecule design.
"Through this effort, we were able to significantly speed up our investigation," Olson notes. "Without the computational power of World Community Grid, it would have taken us many more years to get to this important step in our research."
According to Viktors Bertsis, an IBM lead scientist, the World Community Grid is a global humanitarian effort that has 1.7 million PCs and business computers in more than 80 countries linked to form a giant massive supercomputer, equivalent to a Top-10 supercomputer, to work on some of the world's most challenging problems.
"It crunches data when people aren't using their computers," Bertsis says. "Volunteers install a small application on their computers, and it operates on the lowest priority, using what processing power the users don't need to add to the overall power of the grid. We don't even ask volunteers to keep their computers on all the time; just to give us their idle time. But when those computers are on, it gives a great deal of processing power to researchers who make use of our grid, and it makes good use of energy that would otherwise be wasted. "
Scripps had approached IBM's Corporate Citizenship division, which operates the World Community Grid, because its own grid had perhaps 10,000 people signed up to it and the Scripps researchers were struggling not just with their research but also the challenges of recruiting more people to join the grid and managing the software and systems that kept it going.
"It was taking time away from their research, to the point they felt like they were becoming computer scientists, instead of research scientists accessing computing power, and that was to the detriment of their very important work," Bertsis says.
For IBM's part, it was happy not only to give Scripps access to a larger grid, but also add Scripps' grid volunteers to its own pool.
"We were able to do hundreds of times more computational work for Scripps than they had thought possible," Bertsis notes. "One of the challenges is to get researchers who come to us to think about what they would do if they had a ridiculously large amount of computer time, because even those who have access to supercomputers don't get to access them as often as they'd like. They often don't appreciate the potential until they see it in action. Scripps, as well regarded and as well-equipped as they are, couldn't have accessed a supercomputer 24/7 for five years straight. But they could with this grid."
Olson says that probably 90 percent of Scripps' HIV/AIDS computing in his lab has been done via the World Community Grid and, "in economic terms, the electricity alone that we don't have to pay for to run such calculations is as good or maybe even better to us as if IBM had simply donated money outright."
Bertsis notes that IBM doesn't set any arbitrary time limits on researchers using the grid. Some projects, he notes, are short-term or have defined end point. Others may take years and have indefinite end points. In the end, he says, "as long as they believe and we believe they are producing useful research, we continue to let them use the grid as needed."
IBM donates supercomputer to Rice University scientists
HOUSTON—IBM also announced in March that it will make a $7.6 donation to Rice University for advanced biomedical research. The centerpiece of the donation is an IBM POWER7-based supercomputer that Rice scientists will use in collaboration with the Texas Medical Center to study cancer, AIDS and other complex diseases.
The award is the largest for high-performance computing infrastructure in Rice's history. The supercomputer, called "BlueBioU," is the first deployment of IBM's recently released POWER7 microprocessor, which is capable of 18.8 teraflops—or 18.8 trillion floating point calculations per second—and is as powerful as the combined total of Rice's existing supercomputers.
The system will be used for genomic sequencing, protein folding, drug modeling and simulations of molecular-level interactions in tissues. In addition, the real-time analytics capabilities of the POWER7 hardware and software are well suited for mining vast genomic and medical databases for clues to new treatment options and cures for complex diseases.
Baylor College of Medicine, a Texas Medical Center partner that is collaborating with Rice on the BlueBioU project, is planning to explore the entire spectrum of genomic change in cancer through the application of genome analysis technologies, including large-scale genome sequencing.
The data center is connected to the Rice campus and to Texas Medical Center partners via a new $22 million network with a multi-gigabit backbone and more than a terabit of aggregate bandwidth. In addition, Rice has a new high-availability storage infrastructure that provides multiple terabytes of data storage in the data center.