Cray, ZIB & Bielefeld Univ Team On Advanced Bioinformatics Software

SEATTLE, WA -- Cray Inc. (Nasdaq:CRAY) today announced it is collaborating with Germany's Konrad-Zuse-Zentrum fur Informationstechnik Berlin (ZIB) and Bielefeld University to develop advanced software solutions for tackling some of the most challenging problems in the burgeoning new field of bioinformatics: EST clustering, sensitive protein database searches, and three-dimensional protein structure prediction. Better solutions for these problems could accelerate understanding of the genetic bases of diseases and lead to more-effective drugs and medical treatments. The two-year agreement, effective immediately, will make Cray supercomputer xpertise and advanced technologies accessible to collaborating scientists to Bielefeld University's Institute of Bioinformatics and ZIB's Division of computer Science. The accessible technologies will include the current-generation Cray SV1(tm) and next-generation Cray SV2(tm) (code name) supercomputers, both employing powerful special features originally developed for the intelligence community-as well as cluster systems based on standard microprocessors. Initial results are expected by the end of 2002. According to Bart Mellenbergh, Cray's bioinformatics practice leader for Europe, the Middle East and Africa, the agreement covers three main initiatives: * New algorithms (mathematical formulas) for EST clustering. A team led by Dr. Jens Stoye, head of Genome Informatics (Genominformatik) at Bielefeld University, will collaborate in creating new algorithms for EST clustering. EST (Expressed Sequence Tag) clustering is an established method for determining which genes are expressed in a given cell. The targeted algorithms will address exclusion methods (which genes do not belong to the cell) and verification of non-excluded genes-a task usually done today with some variant of a Smith-Waterman algorithm. * New algorithms for sequence alignment. Determining the function of proteins is one of the most fundamental challenges in genomic research, and function is related to structure. If information on a protein's structure is unavailable, as is often the case, comparisons are made with members of the same protein family. The new "jumping alignment" method has been proven superior to established methods, but its computational complexity has prohibited its use in large-scale database searches. Dr. Stoye will pursue the development of new algorithms that make the jumping alignment method practical by exploiting the special hardware features residing in Cray SV-series supercomputers. * New algorithms for three-dimensional protein structure prediction. There is still a huge gap between the number of known protein (DNA) sequences and the number of known 3-D structures of proteins. Accurate 3-D structure prediction of proteins based on sequence information is one of the "grand challenge" problems of bioinformatics. Dr. Thomas Steinke of ZIB's Division of Computer Science will lead the initiative to create more-effective algorithms that exploit information from atomistic (molecular dynamics) simulations using classical force fields and/or improved potentials based on quantum chemical data. Once again, the special capabilities of the Cray SV-series supercomputers are expected to make the development and use of these algorithms feasible. "Validation and benchmarking of algorithmic approaches under development for the simulation of biomolecular systems require a well-balanced computer architecture. Especially, the expected sustained processor-memory bandwidth on the Cray SV2(tm) system is indispensable for our planned simulations. We are highly pleased and looking forward to making a substantial step towards more complex molecular scenarios," said Dr. Steinke. According to Dr. Stoye, "This collaboration gives us the unique possibility to apply our algorithms in a high-end computational environment. We expect this to result in more powerful tools for the practical bioinformatics community." "Jens Stoye and his Bielefeld University colleagues have had outstanding success in developing effective new bioinformatics algorithms, with a focus on methods for pairwise and multiple sequence comparisons," said Cray's Mellenbergh. "Thomas Steinke and his group at ZIB have long experience in contributing to the development of quantum chemical production codes, including optimizing and parallelizing applications in the area of molecular simulations." Mellenbergh said ZIB currently operates a 544-processor Cray T3E(tm) massively parallel supercomputer and a 16-processor Cray J90(tm) vector computer. "We are excited to be working with individuals and institutions of this high caliber on initiatives with the potential to benefit science and the quality of human life," said Cray Chairman and CEO James Rottsolk. "Cray is fully committed to the ambitious goals of this collaboration with ZIB and Bielefeld University."