Thursday, October 24, 2002
|
|
|
Researchers Verify the Accuracy of a Protein-Folding Model Based on Distributed ComputingBy BROCK READ
A team of university researchers has verified that a large-scale computer model employing distributed-computing technology can accurately simulate the process of protein folding, in which chain-like protein atoms bend over upon themselves to form three-dimensional structures. The model was run on about 40,000 machines worldwide through distributed computing, a process that takes an amount of work usually reserved for supercomputers and breaks it up into small chunks that personal computers can handle over the Internet. The team, led by Vijay Pande, an assistant professor of chemistry and structural biology at Stanford University, compared its computerized simulation of how a small protein called BBA5 folds with a simultaneous experimental study led by Martin Gruebele, a professor of chemistry, physics, and biophysics at the University of Illinois at Urbana-Champaign. A report on the concomitant experiments was published online this month by the British journal Nature. The results of the two studies matched, and researchers in Mr. Pande's laboratory believe their model will now be regarded as an invaluable tool for examining both protein folding and diseases that are believed to result from mis-folding, like Alzheimer's disease. "You have to prove that you can come up with computer simulations that are accurate enough to tell you interesting things," says Christopher D. Snow, a graduate student working in Mr. Pande's laboratory. Mr. Pande inaugurated the project in October 2000, inspired by advances in distributed-computing software and the success of the University of California at Berkeley's SETI@Home, in which personal computers that would otherwise be idle sift through radio signals from outer space for evidence of extraterrestrial life. Since Folding@Home was developed, it has been downloaded to about 200,000 computers, some 40,000 of which run the program regularly, according to Mr. Snow. Participants in Folding@Home download software -- which includes a screensaver and a data-analysis program -- at the project's Web site. When computers running the software have been idle for a certain amount of time, they start to analyze packets of data they have retrieved from the project's site. The packets are sent back to the Folding@Home coordinators when completed. In a typical day, a single computer running the software simulates a billionth of a second of protein folding, recreating the process one atom at a time. It is a deceptively intensive procedure, says Mr. Snow, who estimates that a lone computer would take 2,000 years to complete the analyses. The task would take five uninterrupted years even on Stanford's largest supercomputer. Mr. Snow hopes that Folding@Home increases awareness of a field of biology that might strike people as arcane, though many scientists consider it something of a holy grail. The project "gives a chance for tens of thousands of people all around the world to participate directly in cutting-edge research that can hopefully be applied to medical applications," he says.
Background articles from The Chronicle:
|
|
|
|
 Emory U. suspends professor accused of harassing a law-school employee Bush chooses Dana Gioia, a poet and critic, as NEA head Community colleges play too small a role in teacher education, report concludes Canada's Simon Fraser U. suspends 44 students in plagiarism scandal Grants awarded to 6 international teams for genetics research and training High-speed links connect 2 telescopes 7,000 miles apart -- and the astronomers who use them Researchers verify the accuracy of a protein-folding model based on distributing computing |
Copyright © 2002 by The Chronicle of Higher Education