Calling All PCs

You leave your computer hooked up to the Internet to go make a sandwich, and in your absence your computer engages in a bit of cancer research. Sound absurd? Actually, thousands of home computers today are being tapped to do research, helping biologists process gigabytes of data on genes and proteins. Now anyone with a PC and a modem can take part in solving some of today’s major health problems.

The majority of the world’s 300 million personal computers that are hooked up to the Internet waste most of their time sitting idle. To harness this unused power, researchers are using a system called distributed computing (see “5 Patents to Watch: Collective Computing,”). It divides large projects into many smaller tasks and apportions them to individual computers registered to take part in the project and already loaded with special software. The software crunches numbers either when a screen saver comes on or during the minute pauses between keyboard punches. The computers process the tasks and automatically transmit the results back to a central server for analysis. SETI@home, the University of California, Berkeley-based search for alien life, is one of the best known examples of distributed computing. But now the strategy is not only being used to look for ET; biologists are using it to tackle some of medicine’s toughest jobs.

With these systems, researchers can perform large computations in a fraction of the time it used to take, without buying a $100 million supercomputer. Stanford University chemist Vijay Pande has launched Folding@home, a program that simulates the way proteins self-assemble or “fold” into a three-dimensional shape. Such simulations take immense computational power, but they’re worth it: knowing how a protein folds is key to understanding a number of ailments like mad cow disease and cystic fibrosis, which are believed to result when proteins misfold. Pande’s group is folding an intestinal protein, which assembles in 10,000 nanoseconds. “On a typical computer, you could probably do a nanosecond a day,” says Pande. But with distributed computing, he can simulate the folding in as little as two months instead of 27 years, allowing him to do experiments that otherwise would be impossible.