Technion uses 'sleeping' PCs to speed gene mapping

The gene-mapping system provides results dozens of times faster than previous programs.

technion 88 (photo credit: )
technion 88
(photo credit: )
Using the downtime of more than 3,000 linked computers, researchers at the Technion-Israel Institute of Technology in Haifa have developed a free, on-line system designed to help unravel the mysteries of genetic diseases. The gene-mapping system provides results dozens of times faster than previous programs, providing a fast-forward button in the quest to develop treatments for inherited diseases. The US National Institutes of Health, Harvard Medical School and scientists in France, Canada, Spain, India and Israel have already utilized the system. The findings are reported in the June issue of the American Journal of Human Genetics. Dubbed Superlink-Online, the system reduces the time-consuming process of identifying the exact location of a disease gene in affected families' genomes - a crucial step in developing effective treatments. "Superlink-Online makes feasible some computations that were not previously possible," said Dr. Alejandro Sch ffer, staff scientist at the US National Institutes of Health's Bethesda, Maryland-based National Center for Biotechnology Information. Technion computer science Prof. Dan Geiger and Dr. Ma'ayan Fishelson, who was studying for her doctorate at the time, began developing the system five years. Now, Superlink-Online is running in parallel on 200 computers at the Technion and 3,000 at the University of Wisconsin-Madison. The system can be accessed freely as a password-protected service via the Internet, and the output is combined as if it came from a single computer. "Over the last half year, dozens of geneticists around the world have used Superlink-Online, and thousands of runs - totaling 70 computer years - have been recorded," said Prof. Assaf Schuster, head of the Technion's distributed systems laboratory, which developed the computational infrastructure. Technion doctoral student Mark Silberstein, the system's current developer, said its already formidable power - it recently completed in seven hours of computation what would have taken a full year on a single computer - will be increased dramatically in the near future when it is connected to thousands of additional computers using EGEE, a large European network.