He's devised a way to identify potential drug compounds by using a network of more than 150,000 home computers and some innovative algorithms. He said the method accurately predicts how well molecules will bind to a given protein. Proteins are the ubiquitous workhorses of living systems and most diseases can be traced to protein malfunctions of one kind or another, so designing a compound that binds to a particular protein is an early step in drug development.

Pande will present his method Aug. 25 at the "High Performance Computing in Computational Chemistry" session at the American Chemical Society's national meeting in Philadelphia.

"For almost 20 years, people have been talking about doing drug design computationally, but the real challenge has been getting sufficient accuracy," Pande said. "Our main goal was to come up with methods to really push that accuracy to the point at which our methods are pharmaceutically useful."

In the past, Pande said, computer predictions of binding strength between molecules and targeted proteins have been off by as much as 4 to 6 kilocalories per mol, rendering them essentially useless. But when he tested his new method by calculating some bonding energies that are already known, the results were accurate to within 1 kilocalorie per mol. "I think we're at the point where pharmaceutical companies start to get interested," he said.

To get those results, he tapped into Folding@Home, a global network of more than 150,000 home computers that run computations in the background, pooling their results via the Internet to create a resource with "supercomputing power greater than all the supercomputing centers combined," in Pande's words. He set up the network in 2000 to study protein folding and needed its power for this experiment because accurately predicting bonding energy requires "sampling" multiple conformations of a protein, a computationally demanding process. He also developed algorithms that would enable the processors to work together efficiently to achieve a common goal. Pande said this distributed-computing approach could be used to design new classes of antibiotics. And, as part of a current Folding@Home calculation on a protein critical to Alzheimer's development, he hopes to identify molecules that would bind to the protein, pointing the way toward possible treatments.

Few researchers have a resource like Folding@Home at their fingertips, although some other projects (such as SETI@home, which searches for extraterrestrial intelligence) are using the power of distributed computing. But Pande said his method could still have broad applications. The benefits of speeding up drug development could easily outweigh the cost of a multimillion-dollar supercomputer to a pharmaceutical company, he said. Also, several pharmaceutical companies are already harnessing the computers within their organization, much as Folding@Home does on a worldwide scale. As for academics, their time will come. "One way to think of Folding@Home is as a time machine where we can do the sort of computational work now that would be very easy for any researcher to do in perhaps 10 years. And we can develop these methods and test them now," he said.

The method would not just speed up drug development but also could change it fundamentally. Pande said chemists are reluctant to test molecules that are hard to synthesize, but "one of the beautiful things about computational functions is that the synthesis is trivial. And so we can do the hard work – we can study the things that would be hard to investigate just synthetically and then make suggestions for which ones should be followed up. I think it may open the door to a new range of therapeutics that we just can't access very readily right now."

His research was funded by the National Institutes of Health, the National Science Foundation and the Camille and Henry Dreyfus Foundation Inc.

Stanford University Medical Center integrates research, medical education and patient care at its three institutions -- Stanford University School of Medicine, Stanford Hospital & Clinics and Lucile Packard Children's Hospital at Stanford. For more information, please visit the Web site of the medical center's Office of Communication & Public Affairs at http://mednews.stanford.edu.