May 12, 2006 | What does big (and at least one little) biopharma really want from technology providers? That was the theme of an entertaining session at Bio-IT World’s Expo, featuring top IT and informatics executives from Merck, Lilly, Pfizer, Novartis, and Infinity Pharmaceuticals.
Co-moderator James Golden, CTO Life Sciences at SAIC, kicked off the session by saying, “I forget exactly what this panel is called. Let’s just call it an august panel, and depend mostly on questions from the audience.” Given the prominence of panelists, it was a welcome choice.
Questions ranged widely: What are the most difficult challenges in global operations? What should life science technology providers do differently to avoid the disaster that befell many of their predecessors in the 2000 crash? How is the buy-versus-build decision made? There was surprising consensus among panelists on many issues.
“When we started Infinity [in 2000], we had very strong buy orientation. We were going to buy everything we could,” said Andy Palmer of Infinity. “What I’ve found is that most of the requirements driving purchases had a very short half-life. You could buy a cool new product, but the reality was that the experiment that was driving the need for it might become completely irrelevant in six months. That was a challenge.”
Vendors focused on doing one thing well and those able to fit into an open architecture had the best success at Infinity, said Palmer, who singled out Spotfire as a good example. He labeled the lack of “standard, good-quality, third-party LIMS” as appalling.
On the topic of e-lab notebooks, Pfizer’s Kathy Gibson, VP, worldwide research informatics, said: “I am frustrated as hell that we can’t get a good lab notebook. There are a lot of niche players, and they don’t interoperate with each other. It creates a lot of extra work for the informatics organization.” Pfizer, she said, would be willing to work with partners to share the risk and develop “broader, more open” platforms in that space.
Merck has a tendency to build its own tools, said James Swanson, VP, research information services, who has also held a senior IT position at Johnson & Johnson. “At J&J, we bought a lot more than we built. On the Merck side, we have a lot of really good technologists. We’re [used to] building everything because we have people who know how to do that.” That’s not a good enough reason, he said, so he’s trying to change the culture. “My view is if there are mature products in the market that can meet the need, then why are we building it?” said Swanson.
“I do think there are things we still need to build. It may be an integration layer; it may be for things where there is no market place for that. For us to build our own LIMS systems again, I would argue that there are better ways for us to spend that sort of talent,” said Swanson.
Budgeting for post-sale support is a frequently neglected issue, said George Morris, former COO for informatics at Novartis and now a consultant with Salem Associates. He oversaw the IT and informatics build-out at Novartis’ Cambridge, Mass., research headquarters.
“We literally had vendors lined up at the door. What Novartis has done and will continue to evolve is to bring in technologies on a pilot basis and work closely in partnership with the vendors on concept and activity. Novartis is looking more at bottom-line impact whenever possible, but also top-line — so how do technologies affect the pipeline, how do they effect the operational effectiveness and costs.”
It’s all about the data, said Palmer. “What we learned at Infinity in the last six years is, traditionally, technologists spend a lot of time building applications for people when most scientists really don’t want applications, but they do want to manage and maintain their data.”
Better data and tool integration is a pressing necessity, agreed all panelists and emphasized by co-moderator Martin Leach, a principal at Booz Allen Hamilton. But John Reynders, information officer for Lilly Research Laboratories, injected a note of caution.
“One thing that often gets lost is how important it is to integrate at multiple levels. It’s not just how you integrate you data, or integrate your applications, but also how you integrate your presentation. At the end of the day, we can all say, ‘yee haa,’ and meanwhile the scientist is sitting there with 10, 20, or however many different interfaces, Java this and XML that, they don’t think it’s integrated. That’s as inefficient as ever.”