By Mark D. Uehling
January 13, 2003
| In Indianapolis last year, there could have been a tense encounter between two very different groups of people: scientists and IT professionals. The working assumption on the IT side of the table was that all of the available legacy data would have to be incorporated into a new application. Upon hearing that, a scientist simply laughed out loud, defusing the situation. What had been a requirement for access to all the data became an option instead.
Is scientific IT at Eli Lilly and Co. any different? Many industry observers think so and cite a collaborative culture, no small feat at a company with $11 billion in annual sales. Since the arrival of CIO Roy Dunbar more than two years ago, Lilly has made a concerted effort not to align IT staffers on one side of a conference table and scientific researchers on the other.
"The computer sciences and the life sciences are merging," says Bob Oppelt, director of IT for discovery research and pharmaceutical development at Lilly. "The coordination between the two disciplines, and their involvement in projects, has moved from being desirable to being crucial."
Indeed, Lilly seems to make IT a higher priority than its peers do. As Lilly's Vice President Sangtae Kim notes, Lilly added discovery IT to a companywide review process for all therapeutic areas and key functional departments. "This is an important milestone and signifies the arrival of a new level of partnership between IT and research and development," Kim says.
Blending IT and R&D: At Eli Lilly, scientists define the "what," IT decides the "how." In the post-genomic era, "Collaboration will be absolutely critical," says CIO Roy Dunbar.
As Oppelt and Kim describe the IT ecosystem at Lilly, the IT staffers and the bench scientists have a symbiotic relationship. "Scientists define the 'what' of the application — what are the scientific needs, the functionality," Oppelt says. "The IT or computer sciences side defines the 'how' — what are the development tools, how are we going to capture the data, what are the data movement methods we are going to use. Having scientists on these teams, blended with the IT people, helps us get over the humps much quicker."
Ever politic, Oppelt is reluctant to divulge any of those bumps. But he is willing to say the organizational structure at the company has helped it prioritize. "There are no lack of requests from bench scientists," he says. "There is no lack of ideas from the senior executives. There are a lot of really good ideas that we don't implement because they don't fit well into what we're doing or the timing isn't right."
Oppelt and five other senior research and development executives at Lilly chart the IT course, meeting every two months. On a monthly basis, another committee of seven scientists and five IT staffers implements that strategy. Says Oppelt: "That group will track projects and remove any roadblocks the project team comes across. My interaction with other places leads me to believe this is innovative — that it is this blend that is allowing us to produce at the level we're able to produce."
The organizational structure, in turn, enables the company to give Lilly's scientists computational tools and resources that other companies have struggled to create. The company's Structure Based Drug Design software suite is one example. Lilly had been trying to create something like it for years. Oppelt says the old system would have required a scientist to sit down with a modeling expert and have a customized model created from scratch. The new system lives on a Linux cluster and allows molecules to be entered into the system and easily searched for their applicability against a number of biological targets.
Behind the scenes, Oracle provides the horsepower. Some descriptors — any defined characteristic of a compound, such as its molecular weight — are available immediately. "There are certain descriptors it is easier to calculate on the fly," Oppelt says. "There are certain descriptors it takes a while to produce, and we just house those."
The key is the software's ability to tap into experimental knowledge or scientific expertise that Lilly already has. "The software allows us to use predictive models that would suggest which of these molecules has a higher probability of success," Oppelt says, "and which of these molecules do we really want to stay away from based upon negative side effects or data we have previously produced in toxicology studies."
Flexibility for the scientists is built in, he says. "It is very improbable that every scientist on every project uses all of the different algorithms, different tools. The scientists make use of what they believe is necessary for them. [The system] will suggest which ones might be appropriate as you move forward."
Build It or Buy It
In the case of another homegrown Lilly IT tool, Molecule Library, the system allows both scientists and manufacturing types at Lilly to read all the documents about a compound in the company's pipeline. "The time savings that we calculate per search is calculated at two to three hours per search," Oppelt says.
|Eli Lilly at a Glance
|Company founded: 1876
IT staff: 2,700 (600 dedicated to supporting research scientists)
Sales from Prozac (first nine months of 2002): $4.8 billion
Estimated decline in Lilly market value since August 2000 approval of a generic competitor to Prozac: $45 billion
Hot prospects in Lilly's pipeline: Cialis, for male impotence; Forteo, for osteoporosis.
Another example is Lilly's Sample ID program, which is used when a scientist identifies a compound and registers it in the system. Says Oppelt: "The interface and usability from a scientific perspective was something we spent a lot of time with. Nobody likes registering compounds." That chore is tedious and less glamorous than what the chemists like to do, but it is crucial to any future patent claims that Lilly may make about a chemical being invented at the company. It also yokes together two key scientific fraternities. As Oppelt says, "What this has allowed us to do is tie the biological and chemical information," allowing people from both camps to search the same database and get answers in seconds.
Lilly takes a more strict approach when it comes to saving experimental data. "We are fairly rigid about the way we collect the data so that it is usable beyond the individual team," Oppelt says.
The same applies to decisions about whether to build software internally or buy it from vendors. Lilly has an approximately 50/50 split between applications developed commercially and in-house. "We believe this is consistent with our peers," Oppelt says. The company considers the requirements of a particular project and tries to anticipate how much customization an external solution would require. Says Oppelt: "Clearly, we lean toward customizable software that will allow us latitude in the way that we use it within our processes. If not, we build."
For a tech-oriented guy, Oppelt is uncharacteristically quick to say that the source of the company's IT success is not silicon-based but carbon-based. "It's probably more human than it is technology," he allows. As the company recruits new employees, it hunts for candidates who have skills in both areas. "We look for people in our backyard with computer science backgrounds [who] have life science interests or degrees as well. It is a priority."
The same approach applies to picking evangelists who champion new applications within the ranks of Lilly's scientists. "We have champions already in the business, who are talking about the value of the tools, how they can be used," Oppelt says. "The more respected they are within their individual disciplines, the bigger the benefit for us."
And the benefit for the company? Lilly does not believe the payoff from its IT can be quantified, or that it will be easily traced this year or next year to a major drug. But Oppelt and CIO Dunbar do believe the company's strategy will be apparent — both to competitors and to Wall Street.
"We do have some notable successes with Lilly," Dunbar said at the Bio·ITWorld Expo last March. "They come from the consistent processes that we're building." The current pipeline is considered one of the most promising in the industry, and if that trend continues — if Lilly's computer geeks are allowed to share some of the credit — the Lilly culture will be one that other pharmaceutical companies emulate.
"Pharmaceutical success in the post-genomic era will be a product of the integration of systems and data," Dunbar said. "Collaboration will be absolutely critical."
PHOTO CREDIT: PHOTOGRAPH BY STEPHEN HILL