Dec. 2006 / Jan. 2007 | David de Graaf joined Pfizer in March 2005 to spearhead efforts to understand and judiciously spread systems biology (SB) throughout the world’s largest pharmaceutical company. Working from Pfizer’s Research Technology Center (RTC) in Cambridge, Mass., (see “Pfizer’s Pursuit of Technology,” Bio-IT World, April 2006, pp.26-30) de Graaf reports enthusiastic support from top management while promising results from showcase projects have stirred demand beyond his expectations. He spoke with Bio-IT World’s executive editor John Russell about Pfizer’s growing comfort with systems biology.
JR: Give us a retrospective on the past year as Pfizer’s first director of systems biology. What remains to be done?
de Graaf: The past year has been a bit of a whirlwind. I came on board in March 2005, and we reached eight people by the end of the year, so we’ve only been staffed at the initially appropriate levels since the beginning of this year.
We decided we needed to make clear what the RTC could contribute, what our unique niche was. Very quickly we decided that where we wanted to work was in quantitative measurements on biological systems, rather than, for example, on global ‘omics measurements which are often not quantitative. We decided we would work on modeling approaches that were able to simulate biological systems, so these were not inferential models, but forward-simulating models. That limits us somewhat in terms of the biological space we can work in because generally we need to be in areas that are relatively knowledge-rich.
We picked three projects, one that dealt with enlarged signal transduction pathways playing to the strengths of the RTC in terms of being a central focus for kinases [at Pfizer]. We picked one project focused on a new development area for the company, which is moving into biologics, and where we wanted to understand some of the pharmacokinetics of biologics in order to differentiate them from competitor products. [The third] project took a large global look at hepatic-toxicity, trying to optimize both our predictive ability, as well as the number of assays that we ran in order to get to a particular outcome. The ultimate goal there was front-loading hepatic-toxicity [determination] early rather than where we’re catching it right now. We’ve done [all of] this in partnership with Doug Lauffenburger at MIT and Peter Sorger at Harvard Medical School.
JR: How are the projects progressing?
de Graaf: Out of those three projects, we have delivered on the biologics project. We have been able to help steer that program in a slightly different direction than where they thought they were going, and given them some alternatives to a strategy they were considering. That has been impactful. We are now looking at broader opportunities to interact with the biologics portfolio. One of the interesting things, of course, is that biologics come with a couple things. There’s much less pre-candidate research that needs to be done. It’s usually about picking the right target in the right context. And making the definitive agent is not the difficult thing; it’s really scaling that up. That’s exactly the space where we’re helping out. It is understanding potential efficacy and safety issues, and also PK issues.
Then there are early glimpses around hepatic injury. We have made significant strides predicting liver tox for compounds that fell into a class that previously nobody was able to predict. So these are DILI2-type compounds — drug-induced liver injury. DILI2 compounds are essentially compounds that previously fell through the cracks as they went through internal safety screening — not only at Pfizer, but across the industry — and ended up causing liver damage in patients.
JR: Bruce Gomes (head of modeling in de Graaf’s group) said the projects had really excited interest inside Pfizer.
de Graaf: Correct. We are starting to see more and more people come to us and say, ‘Gee, we want to do more of this. Can you do this in a different area of tox? Can we apply this in selecting better candidates in this particular targeting family?’ I must say, [this is] one of my very pleasant surprises over the last year. We, as a group of eight, are way overstretched, and that’s good. We know that there’s a lot of interest and buy-in across the company for these types of approaches.
One of the problems, because we are still relatively small, has been influencing effectively in such a large organization. So although we see major interest, it’s still very spotty, and I know that we haven’t covered the Pfizer universe. That’s one of the challenges for the year to come: How do you now spread the word effectively and make sure that we cross systems biology across the company to be appropriate for those levels of interest?
JR: How do you raise awareness?
de Graaf: If I had an answer, I would be much better off than I am. The practical tactic we’re taking, which may not be optimal but is what we’re doing, is to do a little bit of everything. I encourage my team to publish. This is one of the reasons for having this very open collaboration with MIT, because there’s nothing like peer-reviewed publications to communicate success within the organization. The other thing — rather than to wait for systems biology to deliver and then [articulate a] strategy — is to say, what are the potential strategic benefits now? How could you apply systems biology across a broad portfolio? Rather than try to bite off everything, we’ve decided to, at least in the first order, focus on biologics, and in the second order, definitely other projects as well.
JR: What kinds of researchers get excited by systems biology approaches? Do they share certain attitudes?
de Graaf: Yes, and I think this may also explain some of the receptive environment we’re seeing at Pfizer, and to some degree across the industry. There’s a general feeling we need to do things differently, that we can’t squeeze out more by just doing twice as much work. So everybody knows there are going to be real changes in how we do science. Nobody knows what those changes look like. Some people are more receptive to that idea than others.
There are people who will hunker down and do twice as much work. There are other people who want to do the same amount of work that just want to do it more smartly ... who look at systems biology as a possibility. Just like in any organization, you have the traditionalists within Pfizer, and you have people who are going to push the envelope and look at doing things differently. That level of bravery is often restricted to people at the bench. One of the great things about Pfizer is that we’ve seen that senior management is willing to really consider doing things differently, and that has been a great boon for me, and a great support.
JR: Do you have milestones you’re looking to reach in the coming year?
de Graaf: I do. The projects that we’ve initiated need to fully deliver. We need to go beyond glimpses of hope, clearly. But I have lots of confidence that will happen. My goal for this coming year is not to do systems biology for Pfizer, but to teach systems biology to Pfizer. That’s a dangerous thing to say because lots of these people have been taught lots of things and they’re very well educated. We are talking not about owning systems biology for the company, but trying to implement the process in different places where appropriate.
One of the roles that I see growing for my group at the RTC is where we work more closely together with project teams that own more of the problem. In the first iteration, a lot of the scientific problem definition was by my group at the RTC. We will really start to bring people on board with the integrated approach rather than having some people in informatics sitting off in some corner, somebody doing some modeling and mathematics in clinical, and somebody in a proteomics group measuring something. Get all of those people together in one place and start to provide an integrated capability throughout the company in appropriate areas.
JR: Will you use other systems biology vendors?
de Graaf: Absolutely. One of the nice things about Pfizer is, because of its scale, we have engaged a lot of people in this space. So whether it is BG Medicine or Genstruct or Entelos, or whether it is a company like GNS; all of those companies have been engaged, and sometimes we’ve done projects with them. One of the nice things about systems biology is that, as a process, the tools that these companies provide fit into very particular niches and spaces. We’re not here to say, ‘Oh, we’re working with Genstruct or we’re working with Entelos, we’ve solved all our problems, we’re done with systems biology.’ Actually we want the companies in this space to keep exploring and innovating and moving, and we want to stay close to them and move with them to implement some of their technologies across our portfolio.
JR: When you encounter a roadblock in terms of your work, who do you call among your peers in the industry?
de Graaf: Well, it’s interesting. With the political roadblocks within the organization, my line management has been incredibly supportive. I couldn’t ask for a more supportive set of bosses. And that goes up all the way throughout the chain. So that part, luckily, I can deal with internally. When it comes to technical problems, there are a few people I stay close to. Carolyn Cho and I talk on a fairly regular basis. Carolyn is at Novartis (head of computational systems biology). And this ends up being useful, whether it’s around hiring or a particular technology or particular company, and we’ve actually looked at opportunities to explicitly work together.
Everybody keeps running into the same toxicity and we can’t solve it. Actually putting our heads together and, more importantly, putting our data together may be something that’s worthwhile; we’re exploring that together with the folks at Science Commons right now, (see “The Advantage of a Drug Safety Commons,” Bio-IT World, October 2006, p. 16) as well as the folks at Teranode. I also speak to my old boss at AstraZeneca, Dr. Adriano Henney (director of pathways capability), and to Dave Cook (associate director), who I was very close to there as well. Anything that comes up in text mining or informatics or data mining, I tend to speak to William Hayes (see “Search and Deploy,” Bio-IT World, October 2006, pp24-33) at Biogen Idec, who used to be on my team at AstraZeneca. These are all outstanding people. There are some people who are more on my periphery. And a small company that we love to work with is Numerica Technologies. They helped us out — this is an MIT spin-off — with some very specific issues around large ODE models and how you interpret those.
Email John Russell.
Subscribe to Bio-IT World magazine.