AIC Seminar Series

Linkage analysis is a tool used by geneticists for mapping disease-susceptibility genes in the study of genetic diseases. However such analysis is often beyond the capabilities of a single computer. We present a distributed system called Superlink-Online for computing multipoint LOD scores of large inbred pedigrees.

Superlink-online achieves high performance via parallelization of the algorithms in Superlink, a state-of-the-art serial program for linkage analysis tasks, and through utilization of thousands of resources residing in multiple opportunistic computing environments, aka Grids. Notably, the system is available online, which allows geneticists to perform computationally intensive analyses with no need for either installation of software, or maintenance of a complicated distributed environment.

In this talk I will describe the scheduling system architecture which drives Superlink-online. The main challenges have been to efficiently split large tasks for distributed execution in highly dynamic non-dedicated running environment, and to provide nearly interactive response time for shorter tasks while simultaneously serving massively parallel ones. The system utilizes resources in all the available grids, unifying thousands CPUs over campus grids in the Technion and the University of Wisconsin in Madison, EGEE grids in Europe, and Community Computing Grid Superlink@Technion (via BOINC) . The system is being extensively used by medical centers worldwide. Since January 2006, over 12,000 interactive genetic analysis tasks were performed, utilizing over 240 years of CPU time.

This work has been done as a part of his PhD in the Technion under joint supervision of Prof. Assaf Schuster and Dan Geiger. It has been published in American Journal of Human Genetics and presented at High Performance Distributed Computing conference in 2006.

Mark Silberstein is a PhD student at the CS department in the Technion under the joint supervision of Prof. Assaf Schuster and Dan Geiger. His main research focus has been efficient serial and parallel algorithms for inference in Bayesian networks (in the context of genetic linkage analysis), and their execution in large-scale opportunistic computing environments, aka Grids. He is currently visiting UC Davis, working with Prof. John Owens on the parallelization of Bayesian inference on GPUs.