Winter/Spring 2007

JEFFREY SKOLNICK is a professor in the Georgia Institute of Technology’s School of Biology, director of the Center for the Study of Systems Biology and a Georgia Research Alliance (GRA) Eminent Scholar in computational systems biology. He joined the faculty of Georgia Tech in January 2006 after serving as director of the Buffalo Center of Excellence in Bioinformatics at the State University of New York in Buffalo.

photo by Gary Meek Jeffrey Skolnick is director of the Center for the Study of Systems Biology and a Georgia Research Alliance (GRA) Eminent Scholar. (300-dpi JPEG version)

Q: What is computational systems biology?
The traditional way people study biological systems is a very reductionist approach. You have a very complicated system and you want to take it apart and study one molecule at a time. You study one molecule or family of molecules and become a world expert on it. That’s very powerful, but it’s like taking a car apart, dropping the parts on a desk and looking at one nut. You pick it up and have no clue where it fits in the car. It has some interesting properties, but the context in which it appears – its function – is lost if you study this one nut.

In systems biology, the goal is to study how the parts interact. If you look inside an individual cell, it’s like a crowded party on New Year’s Eve, rather than an isolated person on a deserted island. The reductionist approach to biology tells us about the isolated person, but it misses the collective behavior. Systems biology attempts to understand and use the collective behavior of molecules to gain greater biological insight.

The long-term goal, whether it’s reached experimentally or computationally, is to understand how things work synergistically.

We develop and apply computational algorithms to predict and understand the function of proteins – the molecular machines of the body, the engines, the workhorses that regulate and control processes. It’s estimated that in the human genome there are 20,000 or so types of proteins, not counting the so-called splice variations that bring that number up to hundreds of thousands. The function of less than half of these proteins is known. If I’m an experimentalist, I want to know which of these proteins to study first because I want to study something interesting and potentially important. So we develop computational methods to prioritize these things.

On a practical level, we’re developing approaches to increase the efficiency of drug discovery. Drug design is extremely important and difficult. We’re very complicated living machines, and drugs can have a lot of side effects for us. We want to adapt the systems biology approach to figure out, first of all, what makes a protein an interesting target of a drug. Then we’ll use this insight and systems approach to accelerate drug discovery. In particular, we’re working with Professor John McDonald, chair of the Georgia Tech School of Biology, on drug targets for treating ovarian cancer.

Computational systems biology is very computer-science-oriented. We use the principles of physics and chemistry to build computer models. We try to reverse engineer – guess the relevant facts, build a system, test it, go away frustrated and then do it again. Occasionally on a good day, we’re less frustrated. I earn my living on the one in 20 days when something works.

Q: What is the potential scientific impact of this field of study?
The potential for scientific impact, in principle, is the understanding of how life works. Let me go back to my analogy of the isolated person on the deserted island versus the crowd at the New Year’s Eve party in Times Square. If you look at the individual components, one molecule is not a living thing. But when you put them together and somehow it works, the whole is greater than the sum of the parts.

We want to understand on a functional level what are the key issues of life that could be controlled or manipulated to allow us to do a lot of engineering once we understand the design principles. Then the implications are better diagnostics and treatment of diseases like cancer and diabetes. It’s like mom, dad and apple pie. It sounds very good, but it’s not happening tomorrow. In the long-term horizon, systems biology is where the revolution in biology is occurring.

Q: What is the focus of your research in this field?
We’re developing general principles and understanding that we’re applying to important problems. One thing we’re trying to do is accelerate drug discovery, especially for cancer, diabetes and inflammatory disease. These are important problems, and we’ve come to the conclusion that biological systems are more linked that we probably thought. So we have to leverage ideas.

Diagnostics are easier to develop. If a blood test doesn’t work, it won’t kill somebody. Developing treatments is harder. That will come later.

Q: How will the center you are building lead to better and healthier lives for people?
You have to take the long-term horizon. It’s not happening tomorrow. It could happen in five to 10 years. We’re developing more powerful diagnostics, particularly for ovarian cancer in collaboration with Professor John McDonald (chair of the School of Biology). We’re developing an approach for diagnosis of ovarian cancer that could be used for detecting other types of cancer.

With regard to drug discovery, if we succeed, we will produce robust predictions for drug targets and reduce the current failure rate of 99 percent in drug discovery. The failure rate is so high because the human system is very complicated. If we’re very lucky, we’ll discover a new drug. If we’re not as lucky, but still lucky enough, we’ll develop a diagnostic.

In science, you have to understand there is serendipity. You never know how things are going to turn out. It’s like you don’t base your financial future on finding the Hope Diamond, but if you step on it, you ought to pick it up.

photo by Gary Meek Jeffrey Skolnick is director of the Center for the Study of Systems Biology and a Georgia Research Alliance (GRA) Eminent Scholar. His work is the marriage of biology and computer science. (300-dpi JPEG version)

Q: What are some research milestones you and your team have recently reached?
We’re very interested in proteins – their structure and biologically active shape. We’ve been curious about how many protein structures exist. We’ve recently published a paper in the Proceedings of the National Academy of Sciences (PNAS) that shows these structures arise from some remarkably simple design principles.

In another study with a talented graduate student named Jake Boggan, we’ve looked at the uniqueness of cellular pathways, which are like assembly lines. Our work shows that these pathways are not as uniquely defined as people thought.

We’re also asking what makes a protein a drug target. What characteristics can we learn from known drug targets that we can use to generalize about the unknown ones? There are only 500 or so proteins that are known drug targets. In this field, you proceed by analogy because the systems are so complicated.

Q: A systems biologist takes a non-traditional approach to creating new drugs. Describe that process and how it differs from the typical approach.
By one means or another in the traditional process, you identify a protein target believed to be responsible for a disease. Then you screen it against a large library of small-molecule compounds to identify binders. Basically, there is a functional region that you want to plug up like a cork. It fits. Then you do medicinal chemistry to optimize it because you want it to fit tightly. You want to enhance the specificity. You don’t want it to bind to something it doesn’t belong with….

Then you feed the drug to an animal model, and it may work in the animal model, but not in humans. You have to deal with absorption and toxicity. There may be side effects. You may have to go back and optimize again, maybe make a subtle change that can have a big biological effect. But back to the analogy of the crowded party on New Year’s Eve. With the traditional approach to drug discovery, at most, you’ve interacted with one family in Times Square.

Systems biology wants to design in drug specificity to look at the biological context, the cellular pathways, the processes that the drug target is associated with. For example, imagine yourself in a city that is separated by a river. There is one bridge across the river. The best way to inhibit traffic out of the city is to blow up the bridge. But if there are 500 bridges, you’ve blown up one. So what? The traffic works around the bridge that is blown up.

Systems biology looks at the characteristics of where proteins appear in the cellular pathway. Is it at a critical point like the single bridge in the city, or are there many interconnected ways? Is it a unique molecule in an organism? Is it used only in this pathway, or are there 50 others? You have to look at the context in which it occurs. The goal and hope – it remains to be demonstrated – is that we can minimize the side effects and enhance the specificity of the drug. The burden of proof is on the field to prove this approach is a better mousetrap.

Q: Why are the state of Georgia and Georgia Tech the right places to do your research?
The intellectual environment of Georgia Tech and Georgia make it the right place to do research. Georgia Tech is a highly collaborative environment, which is very important because systems biology is inherently multidisciplinary. At the core, you need strong experimental biologists. You also need chemists and chemical engineers, as well as computer scientists. Systems biology is extraordinarily data intensive. You need biomedical and other types of engineers, and Georgia Tech is one of the world’s best engineering schools.

In this collaborative environment, I have the opportunity to collaborate with other experts, such as Professor George Nemhauser, who does operations research in the Stewart School of Industrial and Systems Engineering. Here, there are people who have the expertise I lack, but together we can do things we couldn’t do in isolation.

If you want to build a program in systems biology, you want to be in a place that has an established track record in engineering, as Georgia Tech has. You also want to be in a place that understands and values excellence and creativity, and Georgia Tech does.

As for Atlanta, it’s a very exciting and dynamic place. There’s something for everyone. You feel a vitality in the air. And the weather is very nice. Also, Georgia is a remarkably forward-looking state with respect to education. You have the Hope scholarship program, for example.

Georgia Tech has aspirations to be among the world’s best. A lot of institutions say they want to be the best, but have no clue how to go about it. Georgia Tech has a clue, which is essential. It has the resources and the vision.

Q: What was the role of the Georgia Research Alliance in your recruitment?
The Georgia Research Alliance (GRA) enables universities to leverage resources to get the best people, which is important because the best people are expensive. In my research, you need the computational tools, and they are expensive. GRA provided the funds to buy the supercomputer we have in our lab. If you want to hire world-class people, but you don’t give them the tools they need to succeed, then you’re wasting your money.

GRA also provides a significant portion of the support of my endowed chair. They made it attractive for me to come to Georgia Tech. Also, I was able to bring 16 members of my research and administrative staff with me.

Q: What are the plans for the growth of the Center for the Study of Systems Biology?
We want to be one of the world’s best centers in systems biology. So we have to recruit top-tier people, particularly experimental biologists. That is our weakness at the moment, but Professor John McDonald (chair of the School of Biology) is leading the effort to recruit these people. We need 10 world-class researchers doing cutting-edge experiments.

We hope to recruit these people within the next five years, but it will take a lot of money. If we create an exciting environment, they will come. We need these experimentalists to have a world-class program in systems biology. I’m glad to say the Georgia Tech administration is behind this initiative.

Q: How does your research complement the nanotechnology and bioengineering research under way at Georgia Tech?
I recently went to see Professor Jim Meindl’s operation at Georgia Tech’s Nanotechnology Research Center. I was flabbergasted. It is remarkably collaborative. We are now collaborating with them to develop nano-biotechnology. For example, we’d like to have a better way to do DNA sequencing. Maybe we can develop a $1,000 human genome on a chip, and then we can look at the genetic differences between you and me. We’d also like to build a synthetic kidney on a chip, which is nano-bioengineering.

Already, we are leveraging some of our insights into cellular pathways and computational methods and applying them to drug discovery…. This kind of work is essential to the future of my research.

The spinoffs of our ideas and insights will come because of the people who have expertise in bioengineering and nanotechnology. Without these people, these ideas will just remain glints in my eye.

At Georgia Tech, we have people who can build novel devices for detection, control and manipulation. They have the bioengineering expertise that will enable us to do cutting-edge systems biology. Without this engineering infrastructure, we wouldn’t be able to compete. I’m talking about state-of-the-art devices that only a few places in the world can build, and we can do it here at Georgia Tech.

Q: What is the potential economic impact of computational systems biology – particularly in Georgia?
You have to take a short-, medium- and long-term view. When you look at the short-term, I have aspirations to start a small software company to take advantage of the ideas and insights we’re gaining.

More globally, we want to leverage our ideas to accelerate drug discovery, to prioritize drug targets. Computational systems biology can tell experimental biologists to tweak this compound here and there, and that can accelerate the drug discovery process. If you get a $1 billion drug, that’s all you have to do to get a very large economic impact.

If we develop a diagnostic or novel treatment, there will also be a potential economic impact – probably in the intermediate term, not tomorrow. The bigger impact will be over the intermediate to long term, probably in five to 10 years.

Q: Describe the capabilities of the supercomputing power – funded by the Georgia Research Alliance and IBM -- available in your lab.
It’s like having 4,000 computers. The supercomputer weighs 35 tons, requires 80 tons of air conditioning, consumes about 230 kilowatts of power. It’s the sixth largest academic computer in the United States. It’s the largest in the world devoted to a single effort in systems biology. It’s the 51st largest supercomputer in the world by some measure.

And it’s going to get bigger. We’re in the process of benchmarking for an upgrade to increase our capacity by 20 percent. Right now, the supercomputer is always full, though the data – 20 terabytes of it – is always in flux.

Our supercomputer can perform 15 trillion operations per second on a good day. Without this tool, I would just stay home and play with my 2-year-old triplets all day. As an example of the kind of calculations this supercomputer can do, we study a large class of molecules that is a major drug target. We have an algorithm that seems to be among the world’s most accurate at predicting the structures and features of these molecules. With one computer, it would take 70,000 days – that’s 200 years – to do this one calculation. If I had to wait that long, I’d be dead. People wouldn’t care; they would solve the problem another way. On the other hand, with the equivalent of 4,000 computers working on this calculation, it takes just 18 days.

There are thousands of protein sequences and hundreds of genomes we want to understand. Without this supercomputer, we could not do anything. A lot of what we do doesn’t work, so mostly we’re developing computational methods right now instead of applications. We spend a significant amount of computer time testing ideas. Does this work? No. Let’s try this. We’re trying to reverse engineer a very complicated system. You need a quick turnaround time. It cannot take five years for me to find out that my brilliant idea is not a brilliant idea. I want to know the answer yesterday.

You have to do a lot of iterations to learn what works and what doesn’t. We’re dealing with a complicated system. This tool teaches us quickly. It helps us apply what we learn on a large scale so we can figure out whole cellular compartments with tens of thousands of molecules by using a collection of multi-scale molecular approaches. And this is not even close to enough, but we do little pieces of it. You get a proof of principle, and then you let the thing loose.

Q: How do you anticipate that this supercomputing power will contribute to the development of computational systems biology?
We want to know the functions of molecules. The supercomputer will help accelerate our ability to infer the function of a large class of molecules. We’re developing state-of-the-art algorithms to push the frontier forward. Without this tool, we can’t explore new ideas.

The supercomputer allows us to ask a lot of “what-if” questions, which is essential. If research is what I’m doing when I don’t know what I’m doing, then I have to have a lot of time to play. The unanticipated consequences are the most fun.

Q: What is the key to successfully combining scholarly research with entrepreneurship, as you have done?
Partly, it’s luck, and partly, it’s the environment. You have to learn how to do this. When I was in graduate school, I didn’t think of starting a company. Then I spent 10 years at the Scripps Research Institute, which is a very entrepreneurial environment. I watched people take advantage of their ideas.

It also depends upon the field you’re working in and how easy it is to commercialize. There are certain areas of scholarly research that are very important, and who’s to say that in 100 years there won’t be tremendous economic impact. Look at the electronics industry. People were studying spectral lines 100 to 150 years ago. It wasn’t obvious it would lead to the electronics industry. So you have to promote core research, even when you don’t have any clue whether it’s going to yield anything.

It’s nice that in computational systems biology there are practical implications. One of the challenges is developing ideas and seeing them applied so they can have an impact beyond academia. It’s a different challenge, but no less of a challenge, to be part of a successful company. I like hard challenges.

The problem is that every scientist thinks he or she is a business person, just like every scientist thinks he or she is an English professor. A lot of us are neither. So you have to surround yourself with people who are great at what they do so you can be great at what you do. Part of the key is knowing when to hand off, knowing what you don’t know.

Q: How did your interests and opportunities lead to where you are today?
I don’t like to work on what everybody else is working on. You have to have a new and original idea. If you don’t have one in a field that’s already established, why bother? If I have an idea, I’m fearless to pursue it.

For example, when I looked at Professor Jim Meindl’s work on chips, it was so cool. I have no shame to admit that I don’t have the slightest clue about how his group is doing this research in nanotechnology. So I said, “Why don’t we work on building a synthetic kidney on a chip?” He liked the idea, so we’re giving it a try.

I change fields every five years or so because I want to be in a field where I can make a contribution. I’ve been studying protein folding now for longer than I expected because it’s a very hard problem. But I like to solve hard problems. I’m sticking with it.

As a child, my parents were very nurturing. I remember my father coming home with tube experiments for me to do and those “how-and-why” science books. My mother encouraged us to be curious. There was nothing we couldn’t do if we tried hard enough. My parents took us to museums such as the Smithsonian. They always encouraged us to be creative and think outside the box.

Rather than forcing extracurricular activities on us, they said we had to be really interested in it. They encouraged us to be interested in the things we were interested in. I played guitar for a while because I wanted to do it…. I was given the opportunity to be creative.

One thing I like to do is build furniture – desks, bookcases, credenzas. I like to work with my hands. That’s ironic because I don’t work with my hands at all in the office. I don’t have much time to do this now, but I like to design furniture and figure out how to put it together. There’s an element of instant gratification in it. So much of my work is non-gratifying. I go home frustrated a lot. But building furniture is tactile and real. It’s very gratifying.

For my children – my 2-year-old triplets – I want them to become what they want to be. I think one of my daughters may become an engineer. My son is already speaking in full sentences. He’s very verbal and analytical. My other daughter is very detail-oriented and intense. We are nurturing their creativity and making learning fun for them.