Nov 15, 2005 | Good communication, blurring cultural boundaries, and strong project governance may be as, if not more, important as sweeping technology solutions when it comes to converging discovery and IT and expediting drug development. That was one of the overriding themes at the inaugural Bridging Discovery and IT conference* organized jointly by Bio•IT World and Cambridge Healthtech Institute (see also A Win-Win Situation, page 6).
A critical challenge for biopharma is to manage attrition — to use a variety of technology solutions such as in silico toxicology, target data fusion, PK/PD modeling, and bioinformatics to pull risk forward, enabling go/no-go decisions to be made earlier in the discovery pipeline. But according to John Reynders, Eli Lilly’s information officer for discovery and development informatics, says: “It’s not science or technology that’s the challenge — it’s the dynamics and people that’s the challenge. We don’t want a bridge so much as a need to blur the boundary.”
Reynders, who held positions at Los Alamos National Laboratory and Celera Genomics before joining Lilly last year, offered seven keys to successfully eliminating the divide between discovery and IT groups: define community; converge objectives; accountability; streamlining; delegate authority; bureaucracy; and building beachheads.
Reynders has created a single, virtual Discovery Informatics group, borne out of biology, IT, chemistry, and ADME/Tox. Strategies are mapped to projects and services, with bonuses tied to project success. Discovery IT tools are streamlined to limit the profusion of different programming languages and software (e.g., .NET). Bureaucracy is minimized by pushing decisions to local informatics systems councils.
As a result, Lilly’s Informatics programs are managed through what Reynders termed a cockpit rather than a dashboard, which benefits developers by adding components to a single tool. The ultimate goal, Reynders says, is not so much “bridging IT and discovery, but IT and our patients.”
Hope and Hype
Rainer Fuchs, Biogen Idec’s vice president for research informatics and operations, noted the early hype that accompanied the bioinformatics industry, followed by the collapse of firms such as Doubletwist and Informax, while other informatics companies such as LION Bioscience are struggling. Fuchs quoted a Gartner study from a couple of years ago, which concluded: “Bioinformatics is at the bottom of the trough of disillusionment.”
While “data deluge requires informatics,” biopharma companies are “not in the software business,” said Fuchs, and measuring ROI is difficult. But he believes that “a modest informatics contribution that helps a research project beats building the next (unused) ‘killer app.’”
Echoing Reynders, Fuchs said that “effective integration is more important than powerful technology,” particularly in areas such as lab automation, workflow management, LIMS support, and registration systems. Building systems is undeniably attractive, but you must ask: Will they come? Does it matter? And will it last? As for data integration, there can be too much of a good thing. Biogen Idec built a comprehensive target annotation system that incorporated genomic, expression, protein-protein data, and more but required too much energy to maintain that could not ultimately be justified.
Fuchs acknowledged that there is “a lot of buzz” surrounding pathways, simulation, and systems biology, “but very little to show for it” so far. A greater priority is to find ways to reuse knowledge over creating new knowledge, although user interests change with time. Governance is a critical part of the equation: Fuchs is a member of both the research and IT teams, helping to bridge the two.
Juswinder Singh, associate director of the computational drug design group, discussed various cheminformatics solutions employed at Biogen Idec. ChemGate, somewhat equivalent to My Yahoo!, stores and distributes information to researchers. Singh says the system has a high uptake, incorporating data from external collaborators, although it does require a dedicated data entry person. Another database, SARTree, presents data on compound core and subcore structures, revealing areas that are saturated with patents as well as competitor activity. The Integrated Virtual Chemistry platform IVY, developed with Accelrys for in silico chemistry applications, is starting to gain popularity.
Anastasia Christianson, director of discovery medicine informatics at AstraZeneca, noted the importance of integrating clinical data into discovery research. Far from a linear pipeline, drug discovery is increasingly a cyclical process, with pharmacogenomic data feeding into new rounds of target selection. This confounds data integration still further, although Christianson noted there is better use of standards in the clinical realm. Cultural issues also need to be addressed, Christianson said, “M.D.s don’t have or give time” easily.
The closing talk, from Millennium Pharmaceuticals’ Stan Letovsky, offered another variation on the blurred boundary theme — this one with computational biologists and chemists linking discovery groups with informatics and IT. Letovsky noted that grandiose aggregation solutions may not make sense for smaller biopharmas, both for economic reasons and the surplus of knowledge available in public sources. While Millennium benefited greatly from the co-development of Ingenuity’s KnowledgeBase, Letovsky offered examples, from siRNA screening to immunofluorescence experiments, where statistical analysis improved data analysis. Current interests include quantifying the value of biomarkers in clinical trials for the FDA.
*Bridging Discovery and IT – Bio-IT World | Cambridge Healthtech Institute Conference; Philadelphia, Sept. 29-30, 2005.