Kovac, who joined IBM in 1983, has held management positions at IBM Research, including head of computational biology, and vice president of technical strategy and division operations. Kovac is a member of the Women in Technology International Hall of Fame, a member-emeritus of the IBM Academy of Technology, and on the Visiting Committee for the National Institute of Standards and Technology. She holds a Ph.D. in Chemistry from the University of Southern California and a B.A. from Oberlin College.
Q: IBM has become a big advocate for the Interoperable Informatics Infrastructure Consortium (I3C), teaming up with the initiative's original founder, Sun Microsystems Inc. Why now, and why is I3C so important?
A: This didn't start in life sciences for IBM. We've been involved with standards groups for years and in all kinds of forums like the Linux world, where IBM was there in the beginning and not late like Sun. We didn't get dragged into it kicking and screaming.
I really feel that this industry is one where there's a lot of fragmentation. I believe the convergence of IT and biology is a powerful tool. But I think there's a whole [technology] layer missing, what I would call a high-level architectural view where collaboration isn't competitive. I3C can provide this. If you have to partner at the algorithm level, you will be stuck in molasses forever.
Q: But hasn't the core infrastructure problem been solved? It seems like IT infrastructure is not a big priority for Big Pharma. They worry more about finding people in informatics.
A: Look, that's just wrong. You're exactly right up to a certain point. There is this whole infrastructure layer that's about hardware and messaging and networking and database strategy. But up on top are all these applications, and what's missing is a middle layer that allows them to talk to one another.
Say I want to pipeline a major experiment consisting of a sequence of experiments using applications A, B, and D, and I want to do it all in silico. I want application B to use that database and application D to use the 150th database. What tool are you going to use? There are a few companies looking at tools to do that, but right now there's no common way to do this.
Next, tell me what the IT infrastructure must do to support that effort. None of these tools work together. None of these databases can talk to one another. Right now this is all done inside a human being. The researchers have to do it, and they have to do it with keystrokes.
Q: What's the answer to that problem? I3C? Standards groups don't necessarily have a great history.
A: There are two ways to think about it. One is the solution could come as a kind of monolithic operating system for informatics. There are companies who say 'We've got a lot applications. We want to make them all work together.' They also say, 'We're going to provide you with this [operating system] but only for our applications. You have to buy our applications because they are the best.'
I think that's bad. For researchers it's really about using the absolute best applications. Our universities are turning out a tremendous number of the most important applications that people are using — there's huge innovation that happens in government and university labs. We need to be able to integrate the applications that come from both public and private sectors.
So the idea of I3C is to make this layer open, and agree on a set of standards. There will have to be a lot of domain specifics to this middleware architecture, probably done as XML vocabulary around particular areas of chemistry and biology and expression data analysis. And the applications will have to become compliant, so it is a little bit of work for the [informatics suppliers], but ultimately there's a value proposition for everybody.
If you don't do that, someone will come along and be the Microsoft of your information — you don't really want that.
Q: Sun may have been late to embrace some standards, but it led the charge for I3C. Does this cause IBM any concern?
A: I3C started late fall/early winter of 2000, and it was started by Sun with a small handful of its close partners — a kind of 'Sun and friends' group. In January of 2001, I was at a BIO Investor conference and Sia Zadeh [group manager in life sciences for Sun] was there.
So at that meeting I went to Sia and said we wanted to join. We could have said we're going to set up a competing consortium with IBM and its friends. I think he was quite taken aback, but about 20 seconds later he said 'Well, that's the right thing to do.' I give him a lot of credit for recognizing that I3C — at the same time we were recognizing the need — had to be the solution, or something very much like it. We'll compete in the marketplace, but this is really in both of our companies' interest.
Q: What about Oracle Corp.? Will it be part of the effort?
A: I can't speak for Oracle. They have certainly been invited and approached. I think Oracle will want to be part of I3C. If we do it right, and we do it well, that will be good for everybody.
Since I3C's formation, Oracle announced a proprietary informatics platform that they call the Oracle framework, so in some senses we're going to compete. It's like saying Microsoft is open. It's open because everyone uses it? Is that how we're going to define open? I'm sure Oracle would say its [framework] is open. The current I3C architecture includes Oracle. If you looked at the demo on the [BIO 2002 show] floor, it ran Oracle.
The fact is, I3C is going to continue to say can we accommodate all relational databases, which includes Oracle. That's really open. Will I fight Oracle on technical issues when they enter the I3C? I'm sure there will be days when we have those debates in the proper open forums. In the end, my folks tell me code wins.
Q: Where is Big Pharma in all of this? I3C critics point out the slowness of Big Pharma companies to embrace I3C.
A: They are a big missing piece. We absolutely must have Big Pharma in I3C. I have approached a number of folks I know at Big Pharma and said, 'You've got to get your butts in here. This is really important. You are the ones that stand to benefit the most from this.'
They're taking a wait-and-see approach. They do that with everything. They are little skeptical because they've seen standards efforts come and go. They say, 'We tried standards before and it never amounted to much. You know we're looking to you because in the information technology world is where you've proven that these kinds of open standards things can work.'
Look at the W3C [World Wide Web Consortium]. It's got every information technology and computing company you can imagine and we all sit down at the table together. Sometimes we fight — and they go on for years — but in the end people decide what they need to decide.
Q: Why is it so hard for Big Pharma to get excited if the benefits are so clear?
A: They say, 'We almost never operate in that model. Intellectual property — our business is founded on protecting it. Proprietary information — we hold it very close. That's our culture, that's our model.' They can see the need. It is really competitive up here [at the application layer] but this middle layer thing is pre-competitive, and they should collaborate on it.
We really need to grow the community. It's a Metcalf Law thing — the more people that are using it the more value it has for the community. I feel pretty passionate about the I3C.
Q: Much of the I3C effort is aimed at providing interoperability between informatics tools. Has IBM developed informatics software, and if so, how will you use it?
A: We don't sell bioinformatics tools. We've had a research group for about 10 years called the computational biology center and those guys are now mostly trained in biology and IT, but in the beginning they were mathematicians who wanted to apply modeling and systems thinking to scientific areas. They collaborated with biologists looking at how to apply computational biology a decade ago.
Two years ago, we decided [convergence] was starting to take off. Some of our computational biologists have done really interesting research and built powerful algorithms. Now we're saying as a company, this is going to be a huge business opportunity, and it's not just about specialized algorithms. It's about everything we do: It's about high-performance computing; it's about data management; it's about databases and integration and services.
So we have chosen as a business strategy to be an infrastructure leader. We have chosen not to commercialize our algorithms directly as IBM because there are a hundred folks out there specializing, and not just companies but also public institutions. We'll never be as smart as all these guys. Yes, we have a Web site with our algorithms, and the terms are noncommercial use. We basically have made these publicly available.
Q: So companies can download IBM's algorithms from your Web site free of charge? Can you tell us what companies are currently using IBM's informatics software?
A: We'll license to our partners and that creates a closer partnership with them, for example LION Biosciences, Accelrys, and Molecular Modeling. I know we have licensed to pharmaceutical companies — which I can't name — on an as-is basis because they shared the software with their partners who will want to commercialize it. We want to be really clear that this is not an IBM-supported product and we're not in the business of doing that.
We may simply take the 'not for commercial use' off the Web site at some point. We don't want people to copy and redistribute it, but the fact is we certainly don't mind if biotech or pharma companies are going to use it to discover drugs.
Q: Why not get into the informatics business if you have the expertise? Isn't IBM missing an opportunity?
A: I just don't think that's a business for IBM to be in. We're not opposed to commercializ-ing those things. But we'll do it through our partners. Think about it like the open source world. Linux is free, but you can also buy a commercial version.
So here is our goal. Right now this is not the kind of open source the Linux world would recognize as fully open source. Algorithms are a little hard to do that way, and operating systems are actually very easy. It takes a different capability and orientation. You have to have the community and be really established, and you have to be able to take people's modifications, which is not something we've done yet, but it could be a possibility.
What we're not going to do is build the IBM suite of applications.
Q: Lately the informatics industry has suffered notable setbacks and Wall Street won't touch it. What's your take on the informatics industry's health?
A: Unquestionably, informatics firms will survive, and there will be a value proposition for informatics. I don't know whether there will be fewer or more companies. Looking at acquisitions — and the number of acquisitions has increased — you can take that to be a kind of consolidation. Some companies went out of business.
People say you can't make money being a tools company. Well, there are actually some companies that make a lot of money being tool companies. Not everybody has to be a drug company, but I also observe that a lot of people who said, 'We're going to build informatics tools and sell them as software,' are now drug companies. We see that morphing of business models, but it almost doesn't matter.
The question to ask is: Will informatics go away? I think we know the answer to that: absolutely, unquestionably no. Will the use of informatics grow? Absolutely. Will applications continue to come from the public sector? Many will come from it, but I think it will require commercially supported software for pharmaceutical companies to build really extensive information-based discovery systems.
PHOTO BY AMY ETRA