The idea of using idle time on personal computers to sift through huge data sets was pioneered by SETI@home , in which more than 2 million computers are processing radio astronomy data in the Search for Extraterrestrial Intelligence. But now medical research is getting into the distributed computing act.
Last month Parabon Computation Inc., a firm based in Fairfax, Virginia, and the National Cancer Institute invited PC users to "Compute Against Cancer." So far, about 5000 people have signed up to help analyze data on how various anticancer drugs affect gene expression in cells. Parabon breaks up large computational problems into smaller tasks and sends them out to computers that have downloaded the proper software. Processing goes on while the computer is idle, and results are sent back to the company's server. "We're doing this gratis to demonstrate our capability to them," says Parabon CEO Steve Armentrout.
There's "a vast reservoir of power" out there waiting to be tapped, says Armentrout. "Modern computers are so fast that even if you're typing as fast as you can, the processor is yawning." But using PCs for a wide variety of computation and modeling jobs has only recently become practical, he says, because it requires a high degree of Internet connectivity and a secure platform such as that offered by Java.
So far, there is at least one other distributed health research project, run by Popular Power in San Francisco; it started business early this year. The company has several thousand PCs testing flu vaccines and vaccination strategies on a model of the human immune system.
Armentrout says when the cancer job is over next month, Parabon will be starting a protein-folding gig for a researcher at the University of Maryland. He sees a huge future in distributed computing: "There are 100 million Internet-connected PCs in the U.S. and 300 million worldwide. There's a whole lot of computation being wasted."
Popular Power