RETAIL

MOUNTAIN VIEW, Calif. -- Using the power of a 40-processor SGI(R) Origin(R) 3800 supercomputer, the School of Computer Science at the University of Waterloo (UW) in Ontario, Canada, has won the prestigious Critical Assessment of Fully Automated Structure Prediction (CAFASP) competition. RAPTOR (Rapid Protein Threading Predictor), a protein-structure-prediction program developed on the university's SGI Origin 3800 supercomputer by graduate student Jinbo Xu under the supervision of Professor Ming Li, was ranked No.1 for fold recognition among non-meta programs by scientists associated with the CAFASP competition. RAPTOR is an application consisting of completely original components. This type of application is considered "non-meta," as distinguished from "meta" programs, which typically combine features of a number of already-existing prediction programs. As is the case at many of the hundreds of universities around the world using SGI(R) products, the supercomputer is part of a shared resource infrastructure, in this case within UW's Faculty of Mathematics, which includes the School of Computer Science. UW researchers, including Xu and Li, utilize this resource for computational projects across many science and mathematics disciplines. "This work, using the compute power of the Origin 3800 system, is part of a major thrust in bioinformatics that we hope will ultimately lead to the eradication of many diseases that have eluded treatment," said UW Dean of Mathematics Alan George. "Protein structure prediction is one of the key problems that need to be solved to move research ahead in these areas. We greatly appreciate the assistance of the Canada Foundation for Innovation, the Ontario Innovation Trust and SGI in making leading-edge computing technology available for this important research." "We are very pleased that this important application development is being performed on our systems," explained SGI Life and Chemical Sciences Marketing Manager Dan Stevens. "While many customers use SGI servers to power their core high-performance computational bioinformatics and chemical modeling environments, we are especially proud to work with students who are pursuing careers in these fields. Future scientific progress can only come from students today who are working to understand and contribute to these complex research fields." The purpose of the CAFASP competition is to evaluate the performance of fully automatic structure prediction programs available to the scientific community. Competitors predict the 3D structure of proteins expressed by a specific set of approximately 60 gene sequences. RAPTOR's protein-structure predictions required more than 600 CPU hours on FLEXOR, an SGI Origin 3800 server with 20GB of memory. "RAPTOR's No. 1 ranking is a very significant achievement," said Li. "Protein structure prediction is a difficult and very important field. Many well-known researchers have been working on this problem throughout their entire careers." More than 100 prediction programs have participated in the biennial CAFASP competition since its inception six years ago. Under Li's guidance, Xu applied a new linear programming approach to protein structure prediction and developed RAPTOR over a 12-month period. Knowledge of 3D protein structure is vital to medical science progress today because it defines the function of each protein. Using low-throughput manual methods, a scientist typically needs six months to determine the structure of a single protein molecule. Such high-throughput methods as RAPTOR are expected to revolutionize this research. UW uses SGI hardware extensively in a variety of research applications across several departments. The university is currently using approximately 30 SGI(R) Origin(R) high-performance servers, 14 SGI(R) Onyx(R) family visualization supercomputers, seven SGI(R) Total Performance RAID systems, and approximately 125 Silicon Graphics(R) Octane(R) and Silicon Graphics(R) O2(R) graphics workstations.