November 15, 2005 | The inaugural Bridging Discovery and IT conference, which took place in Philadelphia in late September, explored critical issues facing discovery, clinical, and IT teams struggling to communicate and collaborate in biopharma organizations (see Bridges and Boundaries, page 30). I cannot recall a conference where delegates ignored the lure of an open bar, so absorbed were they in assorted roundtable discussions. It was, as one attendee remarked, precisely the kind of material and information that Bio•IT World should be devoted to month in, month out. Point taken...
A defining feature of the conference was a series of joint case studies presented by Discovery and IT colleagues from various biopharma companies. Rainer Fuchs (Biogen Idec) and Stan Letovsky (Millennium) cautioned against building knowledge management platforms capable of aggregating disparate sources of information, without necessarily producing tangible benefits to working biologists and chemists. But Johnson & Johnson’s ABCD warehouse (see How to Spell Discovery, June 2004 Bio•IT World) appears to be an exception to that rule.
“With diversity comes entropy,” said Dimitris Agrafiotis, senior research fellow in J&J’s molecular design and informatics group. ABCD combines chemical and biological data in a “beautiful .NET environment,” but Agrafiotis stressed that “chemistry is an integral part of the database, not added on later.” Listing the priorities as performance, aesthetics, and utility, Agrafiotis said it is “conclusively proven that this is a very good investment” — one that will be rolled out to J&J subsidiaries such as Alza, Cetrocor, and Scios. He declined to list specific examples of return-on-investment: “What is it [you love] about your wife?” he asked rhetorically. “You know a good woman when you find one!”
The speed and flexibility of the database was illustrated in a live demonstration by J&J Cancer Therapeutics Research Fellow, Peter Connolly. He said that ABCD has gained more than 400 regular users across eight global sites within the past 12 months.
Amylin, in San Diego, is fueling its pipeline courtesy of a proprietary Phosmol library. “Building and maintaining the library is an intensive task,” said Michael Hanley, vice president of discovery research. The organizational solution, explained Reed Vickerman, was “doing away with organizational charts” and replacing them with “bubbles.” IT staff were housed in discovery space, research clusters housed in the data center, IT staff were made to feel like “discoverers,” while discovery team members were made to feel like part of the business.
The end result is PIMS — the Peptide Inventory Management System — produced in 90 days, on time, under budget, with users and developers working side-by-side. “We’re not a software vendor,” said Vickerman, “but we’re creating interesting IP useful for other companies.”
Another example of embedding IT groups in research labs was provided by Vertex. The goal for the Drug Metabolism and Pharmacokinetics (DMPK) group was to balance data management into DMPK processes. The solution, according to Steven Schmidt, VP information systems, was to build a learning organization with domain experts buffering information experts on one side, IT experts on the other. Among the keys to success were regular communication, frequent scientific process reviews, and having IT/informatics experts serve as both advisors and internal champions. Consultant costs were slashed by more than 50 percent. And as drug discovery support unit director Mary Sherman noted, with more than half the company’s chemists taking a five-day training course, Vertex has observed a 400 percent increase in the use of PK/PD information and a sharp increase in the number of biosamples analyzed.
Merck medicinal chemist Jason Elliott reviewed the co-development and implementation of an e-lab notebook from CambridgeSoft — a three-year process. The demand for implementation was very much a bottom-up process. “Clear project governance and joint ownership is critical to success,” said Elliott. Performance testing was critical because, as Ingrid Akerblom, executive director of research information services, explained, “scientists break stuff.” Other elements, including LIMS, e-signatures, and offline capability, had to be integrated during the design-and-build process. Despite the lengthy time to deployment, initial results suggest a mean time savings of more than 1 hour/week/researcher.
Like the case studies presented at the meeting, Bio•IT World and our co-organizer, Cambridge Healthtech Institute, viewed the conference as a win-win opportunity. We fully intend to put on at least one similar event next year.