The strategic alliance with MDS Proteomics is a “major milestone” in IBM’s aggressive push to gain a foothold in the life sciences computing market, said Anne-Marie Derouault, director of business development and marketing at IBM.
Under the terms of the alliance, IBM will be the preferred provider of hardware, software and services to MDS and the two companies will collaborate on the development of BIND (biomolecular interaction network database), a publicly available database of molecular-level protein interactions.
In addition, IBM will take a minority stake in MDS, the size of which has not been disclosed.
MDS will use three superclusters of IBM eServer systems totaling 700 gigaflops running Linux and Unix to process output from a network of mass spectrometers located in North America and Europe. IBM’s Shark disk and Linear Tape Open storage systems will be used in combination with IBM’s DB2 Universal Database middleware to store, manage, and access MDS’s protein sequence data.
IBM will also customize its DiscoveryLink data integration technology to help MDS researchers access proteomics data from a wider range of heterogeneous sources. IBM hopes to elicit feedback from MDS researchers using its products with the aim of replicating some of the services it develops for other proteomics partners.
IBM’s Blue Gene protein-folding supercomputer has been tapped as one of several research projects on which the two companies may collaborate.
Frank Gleeson, CEO of MDS, said “We don’t have a specific collaboration around Blue Gene, but IBM seems quite convinced that we can have some collateral benefit to them in that area.”
While this isn’t IBM’s first partnership with a proteomics company — it made an equity investment in Structural Bioinformatics in November — it is its first move into the establishment of a proteomics database. IBM will contribute software and hardware to accelerate development of BIND, while MDS will develop software tools to allow the academic community to access and populate the database.
BIND is currently maintained by the Samuel Lunenfeld Research Institute at Mount Sinai Hospital in Toronto. The database architecture was developed by Christopher Hogue, who is now chief information officer at MDS, while he was an assistant professor at the University of Toronto.
Gleeson said that BIND will move to a new site and the Samuel Lunenfeld Research Institute will play a diminishing role in its maintenance and development.
The public version of BIND, pubBIND, will remain free to all users. A database of proprietary information will be established for in-house R&D at MDS.
Gleeson said that BIND contains atomic-level information on protein interactions with any other molecule. “This is a very detailed and precise mapping of molecular interactions,” he said.
MDS plans to leverage the information gained from the project in further corporate partnerships to discover and develop new medicines to treat diseases such as cancer, AIDS, and depression.
Derouault said that IBM sees MDS as “a major player” in proteomics, a strategic area that the company intends to actively target this year. The alliance serves as evidence that “IBM is serious about the proteomics market and aggressively driving alliances,” she said.
“Proteomics poses challenges in terms of complexity of data storage — genomics was big, but proteomics is an order of magnitude bigger — as well as the complexity of the algorithms, and therefore the computing power that you need,” said Derouault.
“We think this is a strategic area for us to play in because we have lots of assets and strengths in those domains,” Derouault added.
Derouault said that proteomics would continue to play an important role in IBM’s life sciences investment strategy.