As the National Cancer Institute's Cancer Biomedical Informatics Grid (caBIG) project moves into the second half of its three-year pilot phase, it is making a few changes in order to attract participants from the commercial sector.
On Sept. 30, caBIG hosted its first Industry Partners Meeting, which was "intended to be a kickoff for the next phase of the program," Peter Covitz, director of core infrastructure at the NCI Center for Bioinformatics, told BioInform last week.
Covitz said that during the first 18 months of the caBIG project, NCI "aimed most of the funding toward groups doing informatics at NCI-designated cancer centers, because we saw that [group] as the core community that we wanted to engage and get organized with respect to the caBIG concept and the program goals, and also to make sure that we leveraged all of the resources that they already had available in their own bioinformatics and medical informatics activities."
"We never intended for all the work in caBIG to be performed purely by academic groups of varying sizes at cancer centers. … We had always intended to start to leverage the talent, expertise, and resources of the private sector as well."
However, he said, "we never intended for all the work in caBIG to be performed purely by academic groups of varying sizes at cancer centers. … We had always intended to start to leverage the talent, expertise, and resources of the private sector as well."
With the pilot phase of the project at the halfway point, NCI is taking the initial steps to do just that. Around 200 people attended the "outreach and educational" industry meeting that was held on the NIH campus in Bethesda, Md., Covitz said, with "diverse" representation from big pharma, mid-sized biotechs, bioinformatics software companies, large IT firms, and systems integrators.
At the meeting, NCI officials outlined several technical and business issues that they had identified as being of particular concern for potential industry collaborators, as well as several modifications in the project's solicitation policy designed to "broaden the participant community," Covitz said.
For the first phase of the project, he said, requests for proposals were targeted primarily at the cancer centers, "and were very broad in what we allowed them to submit." Going forward, however, NCI intends to "invert" that model so that it will be "more precise" in defining specific technology requirements, "but at the same time, we're going to open up the eligible list of bidders to everybody. It won't just be cancer center groups that can bid."
Covitz said that there are "different vehicles" through which industry partners can work with caBIG, including direct contracts with NCI, or subcontracts through Booz Allen Hamilton, the general contractor for the caBIG project. "For the most part," Covitz said, "we're not doing grants in caBIG. It's not really a research program, so basically people have to start thinking in terms of contracts."
Covitz did not provide a timeline for when the next phase of solicitations is expected to begin, but said that they would be released on an "as-needed basis" over the next year and would be announced via the caBIG listserv (https://list.nih.gov/archives/cabig_announce.html).
Overcoming Barriers to Participation
Covitz said that he and his NCI colleagues were able to assuage a number of industry concerns at the meeting regarding technical requirements for caBIG compatibility as well as IP issues.
On the technical side, he said, "I think most people were OK with our approach to compatibility and interoperability, and the reason is that our approach is not very invasive in terms of technology decisions."
Covitz said that caBIG compatibility focuses "on the interfaces and on the syntax and semantics of the data that gets exchanged. So as long as your system produces caBIG-compatible data structures and consumes caBIG-compatible data structures, it's compatible." (Details of caBIG's compatibility guidelines are available at https://cabig.nci.nih.gov/guidelines_documentation/caBIGCompatGuideRev2_final.pdf).
"It does go against the proprietary lock-in business model, and some vendors still pursue that model, and they're not going to be consistent with caBIG program goals."
This approach, while flexible, may not appeal to every potential industry partner, however. "We do demand that you have those interfaces, so closed, monolithic systems with no programming interfaces, and no way to get stuff out in a caBIG-structured way, are not compatible," Covitz said. "It does go against the proprietary lock-in business model, and some vendors still pursue that model, and they're not going to be consistent with caBIG program goals."
On the IP side, Covitz said that while NCI requires all tools developed with caBIG funding to be released under an open-source license, "any product that was built and funded using private capital can be adapted to be caBIG compatible by building some type of appropriate adapter layer, and then the underlying product does not have to be open source because caBIG didn't pay for it." If the adapter layer was built using caBIG funds, however, then that would need to be released under an open-source license.
Some companies are already working with caBIG. In August, Inforsense signed an agreement with the NCI's Core Genotyping Facility under which the company will develop a web portal that will eventually be available to researchers outside the NCI community as part of caBIG [BioInform 09-12-05]. Yike Guo, CEO of InforSense, said his firm "fully appreciates" NCI's open source policy. While the company's InforSense KDE workflow platform is not open source, "it has an open API that allows scientists to plug any code into it," he said. The company isn't opening up the code for its core workflow engine, "but apart from that, all the interfaces are open," he said.
Some in industry are still taking a cautious approach to jumping into the project, however. Teresa Campbell, data-management team leader at Eli Lilly, attended the recent caBIG industry meeting, but declined to comment on whether Eli Lilly intends to participate "because I want to see further as the project evolves." She noted that Lilly is "interested" in the project, "but that's the best I can say at this point."
Tony Kerlavage, senior director of technology solutions at Applied Biosystems, said that caBIG could help "drive the industry toward open standards" because the thousands of researchers funded by NCI grants will have to comply with those standards themselves, which could have a "ripple effect" across the industry. He did not comment on whether or how ABI intends to participate, however.
David Aranow, director of clinical informatics for biobanking informatics provider Ardais, said that there's still a question of whether caBIG is going to be "a good opportunity for small software products and services companies like ours, or if it's fundamentally going to end up precluding us from serving the academic cancer research community." As an example, Aranow said that NCI is funding cancer centers to build applications for the caGrid, "but there has yet been no definition of what the APIs would be, so commercial entities can't really get a sense of what they would need to do to participate in that space."
Nevertheless, he said, Ardais has been an "active participant" in several caBIG workspaces for the last 18 months, and plans to continue its involvement "not just to be aware of what's going on, but also to help to shape — even as sort of an unfundable member, you might say — to shape the content of" data formats and other standards that will impact its business.
Jian Wang, CEO of BioFortis, a bioinformatics startup based in Baltimore, said he sees a number of opportunities for his company within caBIG, but also has a few lingering concerns that were not addressed at the meeting.
"I do think caBIG is a very visionary step toward where the industry needs to go," he said. BioFortis' flagship product, Labmatrix, "provides the glue, the interface to tie the basic research side and the clinical practice side together," he explained, and without an effort like caBIG, "we'd have to do all the work" to enable integration across that gap.
However, Wang said that it may be difficult to build the critical mass necessary to appeal to commercial parties. "It's like the phone network," he said. "The power isn't in each node — the power is in the connectivity, in the network." Just as the "first few families with a phone probably didn't find it very useful," he said, most companies will be unwilling — or unable — to spend the resources necessary to be among the first participants in caBIG.
So far, Wang said, "it's not clear to me exactly what the incentive is to make that investment. If you have 100 nodes in a network, the 101st can take advantage of the others, but if you're the first, there's no advantage."
NCI, he said, "should be doing more to help vendors get over that hump."
Covitz said that caBIG "still is in a pilot phase, so our hope is that we are in fact creating a new market here … that a market will be there for vendors to consider selling into, and they'll have a competitive advantage if their systems are opened up and caBIG-compatible."
However, he noted, "I won't say that necessarily is the case today. We're still building the basic infrastructure."
By the end of the three-year pilot — in another year and a half — vendors "might be getting a pretty clear indication of whether that market is starting to form, so now might be the time for them to at least start considering whether adding caBIG-compatible interfaces to their systems is going to give them a competitive advantage," Covitz said, "but that's for them to determine."
— Bernadette Toner ([email protected])