With new industry partner Sun Microsystems and its funding finally coming through, the Biomolecular Interaction Network Database is pushing ahead with its development, aiming at becoming “the world’s most comprehensive repository for biomolecular interaction data,” according to its latest press release. But it remains to be seen whether proteomics researchers will accept BIND as the standard for protein interaction data.
Last week, the Blueprint Initiative, which administers BIND, said it had finally “secured” CA$29 million ($21.6 million) in funding, which includes previously announced awards (see PM 6-10-02) of CA$12.5 million ($9.2 million) from Genome Canada and CA$5.4 million ($4 million) from the Ontario Government’s Research and Development Challenge Fund, as well as CA$1 million ($0.7 million) from the Canadian Institutes of Health Research.
As it turns out, this funding has only become available now, due to some twists and turns in the industry support for the project. “While the funding was announced earlier, we can now say that we have the cash in hand to operate,” said Chris Hogue, the principal investigator at the Samuel Lunenfeld Research Institute in Toronto who runs Blueprint, in an e-mail interview. This change required industry contributions — making up the remaining CA$10.1 million ($7.5 million) — that come largely from Sun Microsystems, as well as from MDS Proteomics and Foundry Networks. “An in-kind match from Sun and other industrial partners was absolutely required to get the cash flow from our government agencies,” said Hogue.
That’s because IBM, one of BIND’s original partners, backed out. IBM and MDS Proteomics each pledged CA$4.5 million in cash and equipment back in June of 2001 (see PM 1-28-02), but “IBM could not come to terms with its promised contribution,” Hogue commented, and MDS Proteomics — itself not loaded with cash — contributed equipment worth CA$2 million only “up to November 2002.”
Sun stepped in last February, when it delivered more than $5 million worth of hardware, software, and storage equipment to BIND. According to Charles Mair, director of education and research at Sun Microsystems Canada, this equipment, along with a soon-to-arrive Linux cluster, constitutes Sun’s matching contribution. In addition, Sun will make the Blueprint Initiative a “center of excellence,” Mair said — a designation that could lead to collaborative research projects with Sun researchers in areas such as software development.
With funding secured, Blueprint plans to rapidly expand BIND. At the moment, it has a staff of 59, including 12 database curators and 29 software developers; as well as students, administrative staff, and others, and plans to increase its staff to 77 by 2005. Over the last year, database entries have already more than doubled, from about 7,000 interaction, pathway, and molecular complex records to about 18,000, most of these from yeast. By mid-2005, Blueprint wants to get up to 50,000 records, as well as release new software and related tools such as visualization modules.
Up until now, most of the records have come from the literature rather than from direct submissions, but Blueprint is hoping to change this. This week, Blueprint published its database curation standards for BIND in order to demonstrate that by next January, it “will be fully prepared for major funding agencies and journals to make BIND submissions a required condition of publication,” according to its latest press release. So far, only Genome Research requires authors to deposit their interaction data to BIND. Hogue is hoping other journals will come on board but said he had not approached them yet because BIND first has to ensure efficient curation turnaround for direct submissions.
So is BIND on its way to becoming the GenBank for protein and pathway interaction data? “It will take a global reach, long-term sustainable funding, and demonstrated database curation excellence, together with overall “customer” satisfaction,” said Hogue. “Biologists need to be comfortable with using the database, and find it useful for their research.”
Some scientists, though, are reluctant to commit to one database yet. A team led by Marc Vidal, an assistant professor at Harvard Medical School who studies protein networks in C. elegans and yeast, uses a number of interaction databases for its research. These include not only BIND, but also the Database of Interacting Proteins (DIP), which is maintained at the University of California Los Angeles and contains 18,494 interactions to date, as well as the MIPS yeast protein interaction database, maintained in Munich. Moreover, his group finds it easier to download large datasets directly from researchers’ websites, rather than from a database. With regard to DIP and BIND, Vidal said, it’s “still a bit too early” to say which one is better.
While acknowledging the desirability of a central repository that will archive interaction data permanently and may serve as a hub for data submissions, Vidal said he is not sure whether any of the existing databases is up to the task yet. For now, he prefers that interaction databases stay compatible with each other, for example by participating in HUPO’s Proteomics Standard Initiative, which has been working on an XML model for the exchange of protein interaction data (see PM 2-3-03). “In this field, I am not sure yet that anybody can impose their standard,” Vidal said.