For the second installment of Genome Technology’s systems biology roundtable series, we took a practical look at the field. What will make systems biology a reality in academia, in pharma, and as a sustainable discipline seen as scientific rather than conceptual ¯ and how should scientists prepare themselves for it? Our participants — two from IT backgrounds, one from a biotech, and one from a university — met in Boston this May and came up with plenty of suggestions, though they agree that there’s no easy answer. What follows is a transcript of their discussion, edited for space.
GT: Let’s start out with quick introductions to let our readers get to know who you are.
Wade Rogers: I’m president and CEO of a small company in Philadelphia called Cira Discovery Sciences. I’m a physicist by training, and not a biologist. Cira’s interest is in discovery of biomarkers for diagnostics and therapeutics.
Ian Welsford: I run an applications science group at Fujitsu Biosciences. My background is in molecular biophysics and I’ve worked in academia and other private-sector entities. The group that I’m with at Fujitsu has been around for a year and it’s designed essentially to try to bring technology to the North American life sciences sector.
Tim Gardner: I’m an assistant professor at Boston University in biomedical engineering. My research there has been focused on applying computational and engineering methods to defining systems of interacting genes and proteins — mapping and modeling the functional relationships.
Deborah Riley: I run the applications group at Ingenuity Systems. We develop pathways analysis software, so we’re right in the heart of the systems biology field. We work a lot with biologists and pharmaceutical companies as well as bioinformaticians. I’m a molecular biologist by training.
GT: We’re here to talk about how to implement systems biology, so let’s talk about the elements that comprise it. What would you list as the disciplines or backgrounds important for a systems biologist?
Welsford: The obvious place to start is you’d have some sort of bioinformatics and IT infrastructure, and someone to manage that. A molecular biology focus obviously for gene regulation. But I think where a lot of efforts have really seen a lack of participation has been in people who really understand networks of genes working together — people who understand molecular physiology. There’s a lot of people who understand informatics; there’s a lot of people who understand the biophysics and computational chemistry of individual proteins and small molecules; there’s a lot of people who understand pathways. But you have to in my opinion link that in with somebody who understands a little bit about systems physiology and what happens at an organ level or a tissue level in order to build these interacting networks in some kind of meaningful framework.
Gardner: The bigger question for us is asking what do we want a systems approach to do for us. That tends to define what we need in terms of resources or a team. As it stands we have a mixture of computational folks, physics folks, and molecular or cell biology people. From a training perspective people usually specialize in one type of skill — either computational/theoretical or experimental. I always try to get people to spend a little time doing both because it helps being able to communicate and understand what it’s like on the other side. We usually have computational people do three to 12 months in the wet space and vice versa for experimental people. Rudimentary database skills would really dramatically improve what a biologist can do.
Riley: I look at systems biology as the analysis of physiology viewed at different levels — molecules, genes, proteins, small molecules. Working in-house we have bioinformaticians, molecular biologists, and computer scientists.
Rogers: At Cira we build virtual teams — we don’t build actual systems biology teams. A discipline that I think sometimes gets overlooked in the context of systems biology is the clinician, and that’s someone that we ought not to overlook because if you’re talking about the spectrum from the molecule to the physiology, the clinician is the one who has the real bird’s eye view.
Welsford: If you’re going to have halfway decent algorithms, essentially you’re going to have to have people that have a foot in many different camps — that understand, ‘OK, the mathematics of what I’m doing has this biological model in mind.’
Gardner: I struggle with it all the time, and I think the answer is you cannot possibly train true systems biologists who are able to do informatics, databases, algorithm development, theoretical studies, biochemistry, molecular biology experiments, etc. It’s just too big. I’m often in the position because I’m trying to manage those types of people who are trying to understand all these things and I constantly feel that I’m trying to do too many things and understanding none of them.
Rogers: By virtue of the fact that we’re all sitting around this table, I think we have probably all had the common experience that the interesting things to do are the things that are at the junctures of traditional disciplines. I think the thing that characterizes a successful team is not so much the ability for every person to know all about everyone else’s discipline but for every person to have a sufficient interest to at least penetrate the jargon and develop a common language. It’s that ability to communicate across disciplines that is the key.
There’s a difference between the practitioners of systems biology today and the trainees who will be the practitioners tomorrow. I seriously doubt if in 20 years there will be a systems biologist who knows everything that you just described.
GT: How do you facilitate that kind of communication?
Rogers: I don’t know the answer to that question. In my experience you happen across folks that have this common interest. Facilitating it is a difficult thing to do — but maybe it’s an unnecessary thing because people are constantly circulating. Another thing that’s happening is that now that we are in the era of the genome there’s a tremendous recognition that is occurring where traditional biologists or physicists or computer scientists are beginning to realize that there’s something beyond the borders of their discipline. I think that there’s no shortage of people with whom you can strike up that kind of a relationship and build that common language.
Riley: Working with scientists, everybody has their own project, and I think they’ll only make the effort to go out and learn something new when it’s going to help them with their project.
GT: We hear a lot that academia is a tough place for systems biology because people are so frequently moving from one place to another, and that pharma is a tough place because of its silos. Where do you see systems biology gaining ground?
Gardner: Part of the problem is that no one is really clear on the different skill sets that are needed. At BU we have a bioinformatics program and a biomedical engineering program; those are the two most closely related to systems biology. The bioinformatics is a combination of biomedical engineering, biology, chemistry, and physics. On the biomedical engineering side people run the range from molecular biology, genetics, bioinformatics, all the way over to instrumentation and electrophysiology and mechanics. We struggle with the issue of the curriculum all the time. What doesn’t seem to be practical is teaching all of those things to everybody. Maybe there’s a way to train people with a couple of subspecializations which would allow you to build bridges between disciplines. It’s a tough problem.
Rogers: We make an end run around the whole issue. If you can’t create it all in one place, whether it’s a pharmaceutical company or at a university, you can find the pieces of it — they’re out there. That’s what we do: we build partnerships with the pieces that we don’t have ourselves. For example we work with intramural researchers at NIH and we work with academics at UMass medical school and we work with basic researchers at the Wistar Institute in Philadelphia. Sure, it would be great if all of that was in one place, but it’s so tough to achieve that.
Welsford: Fundamentally you have to set realistic goals of what you’re going to be able to achieve in a given academic program. You cannot expect any graduate or undergraduate program to provide comprehensive experience at everything; it’s impossible. The academic programs to a certain extent are driven by market pressure — so if you look at the jobs that are posted and what people are looking for, one of the problems that people who are broadly trained early on in their career have is they can’t get an entry level job. When people are hiring as a manager they think, ‘Sure it would be great if they knew everything but this person has to be able to do these four specific things’ and that generally is a highly silo type of training.
To some extent these systems institutes that are being set up are an attempt to deal with that on a post-graduate internship basis — give people experience in other areas beyond what they’ve gained in their training. It’s unrealistic to expect that to happen early in a career.
Gardner: That brings up an interesting idea that a part of training could be an exchange program — a formal educational experience. It’s hard to do as a postdoc because when you’re applying to get a postdoc, whoever’s hiring you wants you to have skills; they don’t want to train you for a year. Maybe the time to do it is as part of a PhD program — a foreign exchange type thing.
Welsford: Some kind of obligate, out-of-discipline internship is required. That’s a good idea.
Gardner: We do something on a small scale. It’s just started but has already shown great promise in my lab. Everybody has a project — some are related and some aren’t. I set up a system called peer advising, where instead of those people coming to me and telling me what they do, they come to me with one other person who’s not working on their project — not only so they can learn about another project but also to provide external feedback. It’s been extraordinarily helpful already.
Welsford: That actually mirrors a lot of what we see in the pharmaceutical industry where they’re moving away from classic silos into project teams associated around classes of targets.
Riley: In the pharma companies we’ve actually been very encouraged by their adoption across the value chain. We started selling our software more on the target identification and validation side and we actually found people using it in toxicology before we even tried to talk to those people. There are always people who will look for new technology.
Welsford: You have to look at the drug companies and understand that they’re working under an extraordinarily tight set of regulations. They’re acting in a regulatory climate that’s governing what they do. The FDA has to move into the systems biology world and understand the combinatorial-type therapies that are essentially needed. They are moving in that direction, but who wants to be the first one there to test that ground?
GT: Last question: For scientists currently in the field looking to get into systems biology, what are the top things they should be doing to prepare?
Gardner: Maybe sort of having a minor in some orthogonal field. Certainly a degree of computational skills is important. It doesn’t mean you have to be an algorithm developer. But the natural outcome of all this research is structured databases, and even if you’re a biologist it’s going to help you to be able to query those databases.
Welsford: We’re setting up a short-course program where we bring together industry and nonprofit researchers and educators and essentially have a forum [for a couple of days]. It might be useful to have that kind of thing.
Rogers: We actually did that in Philadelphia last year. The Greater Philadelphia Bioinformatics Alliance formed a virtual institute. I taught a unit in that on proteomics. Actually I think it was very successful — there were people who came from industry as well as academia. They’re doing it again this year. That’s exactly the right thing to do as a way of getting all these diverse people to talk to each other. That’s where systems biology is coming from.