Madison, Wis. – A collaborative project between the University of Wisconsin-Madison and the Technion-Israel Institute of Technology in Haifa has shown that thousands of sleeping computers can unravel the twisted code of genetic diseases.
Superlink-Online, a program first developed at Technion for identifying the location of diseased genes in the genomes of affected families, has since grown to operate on 200 computers in Technion and 3,000 at UW-Madison. The system, which was detailed in the June issue of The American Journal of Human Genetics, operates on the Condor high-throughput computing program developed at UW-Madison, and may hold the key to developing treatments for inherited diseases.
Downtime gene mapping
With the use of multiple computers, the Superlink system provides results dozens of times faster than previous computer programs, reducing the time-consuming process of identifying the exact location of a “disease gene” in family genomes.
When a genetic structure is programmed into Superlink for study, the information is divided up into small pieces and distributed between multiple computers that are hooked up to the system. The Condor system manages these computers, making sure they work on the tasks while they are in the middle of their idle cycles.
Once the computers have completed their sections of the problem, the results are gathered and combined as though the calculation was done on a single computer. According to a recent paper published by UW-Madison and Technion, the system utilized about 460,000 Central Processing Unit hours between June and December of 2005, and makes feasible some calculations that used to be impossible.
“We wish to employ millions of computers of ordinary people, whenever their computers sit idle, to help resolve the mysteries of common genetic diseases,” said Dan Geiger, a professor of computer science at Technion and one of the primary developers of Superlink-Online.
The system was first developed by Geiger and Dr. Ma’ayan Fishelson in 2001, and was originally designed for a single-computer platform. Three years ago, Geiger and his team decided to take the success of this platform and pursue what Geiger called a “parallelization” of the program to increase its speed, which led the team to the Condor project.
Miron Livny, a professor of computer sciences at UW-Madison and head of the Condor project, said both groups began to use Superlink-Online for joint research. The system was set up so projects could be submitted through a Web portal at Technion and transferred to UW-Madison, where Superlink computing also can be sent to the National Center for Supercomputing Applications in Illinois.
“We are not involved in the biology part of it, but we are involved in how to run it effectively on a large number of what we call `opportunistic’ computers,” Livny said.
Since its implementation, the Superlink-Online system has been used by the U.S. National Institutes of Health and Harvard Medical School, as well as scientists in France, Canada, Spain, India, and Israel.
Geiger said the Genetic Center in northern Israel is using it to study a rare genetic disease that distorts women’s palms. By studying the defective gene and its relation to a secondary gene, researchers have a better chance to keep other family members healthy.
The Superlink team has been encouraged by the early returns, and it intends to extend the system. Assaf Schuster, a professor of computer sciences at Technion and head of Technion’s Distributed Systems Laboratory, said the system will be extended to more machines and will incorporate a scientific grid infrastructure known as the Enabling Grids for E-sciencE (EGEE), which could open the system to thousands of computers in Europe.
“As far as distributed computing, it is definitely a very demanding application,” Livny said. “We are very far from being done.”