When a patient doesn't respond to typical drug treatments for epilepsy, doctors sometimes need to take a radical approach: They remove the part of the brain that causes the seizures.
"Understanding the pieces is really important, but it doesn't tell you as much as you might like about how everything works altogether." --Irene Eckstrand, NIGMS
Unfortunately, figuring out which part to remove isn't easy, or accurate. Patients are outfitted with electrodes to record brain activity. An epileptologist reads the records to determine where the brain's pathology lies. Decisions are made and the cutting begins. But those decisions are often wrong: About 30% the patients who have their brains resected continue to have seizures.
Mark Kramer, an assistant professor at Boston University, is applying his math and physics training to raise the success rate. Working with a clinical team at Massachusetts General Hospital in Boston, Kramer is using computer algorithms and other data-analysis techniques to help clinicians do a better job figuring out where to cut. "They're collecting vast amounts of data from these electrode grids," Kramer says. "We'd like to develop quantitative techniques that can tease out information that's impossible to see through a visual inspection" of an EEG chart.
Kramer's work is just one example of how the line between physical science and biology is blurring as a growing number of researchers set their sights -- and their mathematical chops -- on hard-to-solve biological problems. The trend is especially apparent in biomedicine, a discipline within which researchers, eager to find practical treatments for disease and to push outward the boundaries of knowledge, are finding new ways to use math to usefully rechannel a flood of data created by new technologies. It's important work, and opportunities are increasing, but the interdisciplinary nature of the work presents some problems that those entering the field should be aware of.
Forests and trees
Also in Science Careers this week:
Informatics Careers Take Shape in Translational and Clinical Research - Electronic patient data and research repositories mean new opportunities in medical informatics.
Statistics Serving Biomedicine - Spanish statistician David Rossell supports other biomedical scientists while pursuing his own research.
Trained as an applied physicist, with additional mathematical training, Kramer says biomedicine is a perfect fit for him. "I think physics [is] one of the best types of training to get for these interdisciplinary fields," he says. "Physicists have a lot of experience in modeling complex systems. And because they understand a lot of the mathematical tools that biologists may not have used up to now, they are often more adept at playing with the mathematics or more comfortable with the computer tools. I'm essentially doing the same thing I always did, except instead of ions or molecules in a gas, it's neurons in the brain."
His more traditional biology colleagues, meanwhile, help to keep his physicist's reductionist tendencies in check; biologists, he says, know where to look for complexity. "They'll also have a better sense for whether a result is reasonable or not, so they're really critical."
Irene Eckstrand of the National Institutes of Health's National Institute of General Medical Sciences, agrees with both ideas: that quantitative expertise is a boon to biomedical science and that the biologist's perspective remains essential. "I think sometimes people in the more quantitative sciences tend to oversimplify the difficulty of biological questions. But they're so complex."
As program director for a national effort to develop computational models of the interactions between infectious agents and their hosts, Eckstrand brings together groups of physicists, mathematicians, computer scientists, and biologists to simulate the spread of infectious diseases such as H1N1.
Eckstrand points out an irony: As biology moves from a reductionist picture of how things work -- one that historically is modeled on physics -- there is an increased need to better understand complex systems. "For a while there, we were deeply into molecular biology, and it was a matter of 'if we understood all the pieces, we would understand how the systems work,' " Eckstrand says. "Understanding the pieces is really important, but it doesn't tell you as much as you might like about how everything works altogether." That, she says, calls for physicists and mathematicians with expertise in complex systems.
Nancy Sung, senior program officer with the Burroughs Wellcome Fund, who oversees the Career Awards at the Scientific Interface, a program that funds researchers trained in the physical, mathematical, and computational sciences to begin working in biology, says the flood of data and advance of technology in biology have made it an attractive place for people in the physical sciences to work. Advances in imaging, for example, have facilitated real-time, noninvasive analysis. [Editor's note: Sung is also the program officer overseeing a grant to Science Careers that funds CTSciNet.]
But Sung insists it's not a one-way street. "Our approach has not been that we have these biological questions and we need technical help to solve them, but that this is an exciting science that can deliver value back into all of these fields," she says. "A physicist should be able to do really good physics here, not just provide technical help to a biologist. It should be a situation where both sides are involved in raising the questions and figuring things out what the directions ought to be."
A chalkboard among the whiteboards
Joshua Plotkin, a theoretical mathematician who works in the biology department at the University of Pennsylvania, is a good example of the sort of cross-pollination Sung advocates. "It's a wonderful two-way collaboration," he says. "I desperately need to interact with experimentalists doing experiments, and very often they find themselves in need of technical advice on how to deal with data sets that they've generated." The two-way communication helps guide and refine his mathematical models, which he designs to elucidate the forces that determine genetic variation in microbes and viruses.
Working as a mathematical biologist in a biology department sets him apart from his colleagues in at least one detectable way. "I'm the only one with a chalkboard," he laughs. "Everyone else has a whiteboard, and I wasn't raised with whiteboard."
Apart from that, he says, the department has been very welcoming of his quantitative techniques. "It seems clear to everyone, not just in this university, that quantitative techniques in biology will be increasingly useful in making progress. I think more and more bread-and-butter biologists find themselves facing a huge amount of data generated by their powerful machines and they don't know how to make use of it all."
Plotkin became interested in evolutionary biology after studying math at Harvard and Princeton universities. "It was a real event for me to realize, having studied pure math for so long, that there's this whole world of evolutionary biology that's just waiting to be understood in a quantitative way," he says. "It's sort of an extra bonus that some of this theory might actually be helpful to human health."
Among Plotkin's interests is the evolution of the influenza A virus. Viruses are subject to the same laws that constrain the evolution of higher organisms, but they evolve much more quickly: the genetic changes in hemagglutinin, the main flu protein, over the past 20 years are equivalent to those occurring in a typical mammalian protein over 20 million years, Plotkin says. And unlike the mammalian case, there are extensive records of flu-virus mutations.
Plotkin is looking at the slight variations that occur from year to year within a single subtype, such as H1N1. Unlike the variations that occur when a virus moves from an animal to humans, which are rare and random occurrences, the annual variation might be predictable and analyzable using mathematical techniques. Plotkin combines pencil-and-paper mathematics, computation, and statistical analysis to those records in an effort to figure out how genetic variation in the virus is related to the structure and function of the protein, including its virulence. "If we can understand that, we can to some degree predict it," he says.
"It's the ability to understand why the flu has changed as it does from year to year, in the ways that it has changed, that might help us figure out where it's going to change next. We can then calibrate our vaccines appropriately," he says.
Finding her place in the job market
Julie Biteen, who joined the chemistry faculty at the University of Michigan in December, says her interest in biology and medicine emerged slowly. She was turned off by undergraduate biology, frustrated by what she considered mounds of memorization and a lack of precision in her biology classes. "It was very cartoony at times, and I felt like there was a lot of hand-waving."
So she pursued a Ph.D. in solid state chemistry at the California Institute of Technology (Caltech) in Pasadena working in the labs of Nathan Lewis and Harry Atwater. "It was a very practical type of thing, but it wasn't a save-the-world kind of application," she says.
Looking to "switch gears," Biteen completed a postdoc at Stanford, doing single-molecule spectroscopy with W. E. Moerner in collaboration with developmental biologist Lucy Shapiro. The move allowed her to continue doing optics -- but instead of characterizing the properties of a sample made up of many molecules, she now worked on single-molecule imaging, looking at the fluorescence emission of one fluorescent-protein molecule at a time.
In her new lab at Michigan, Biteen is working to help biologists track single molecules in living cells. "My contribution has been to help develop techniques that are relevant to live samples and to bacteria cell samples," she says. Finding ways to illuminate such tiny structures could help in the development of new diagnostic procedures for early cancer detection, or aid studies of infectious disease. Right now, she says, there is a lower limit on what biological imaging techniques can see. "But if we could decrease that minimum signal we could find ways to catch cancer earlier."
Just a few months into her new position, Biteen is setting up collaborations across the campus. Because her method allows researchers to view samples that they can't view using other techniques, she says she expects the technique to be "an easy sell" to other researchers.
Selling her skills during the job search was more challenging. The advantage to being interdisciplinary in the job market is that you can fit into different types of departments. But getting hired in those departments can be tricky. Many departments don't yet know what to make of interdisciplinary scientists, Biteen says.
Because the job market was "pretty bleak" last year, she applied for jobs in a number of different disciplines, including chemistry, electrical engineering, and biomedicine. "Some schools looked at my application and said, 'You are clearly not an electrical engineer,' while others said, 'You do optics, of course you're EE,' " she says. "Bioengineering was also a place where some schools differed. Chemistry was always an easy sell because I think chemistry is inherently pretty multidisciplinary."
Learning three different fields
Last fall, Hadley Sikes joined the chemical engineering department at the Massachusetts Institute of Technology (MIT) in Cambridge where she is developing enzymes that can mimic the immune system. She set her sights on engineering while completing her postdoc in physical chemistry at the University of Colorado, using polymerization reactions to detect molecular recognition. "I was drawn to engineering's emphasis on service to society and wanted to take my quantitative skills and apply them to technological developments that will benefit society."
Her background in physical chemistry and kinetics provided the knowledge she needed to analyze redox reactions. But she needed to learn how to change the performance of the enzyme. So she took a second postdoc position, in Frances Arnold's lab at Caltech, and learned to reengineer redox enzymes using "directed evolution."
Now, back at MIT as the Joseph R. Mares Assistant Professor of Chemical Engineering, a job she started last fall, she is developing enzymes to mimic the immune system in hopes of finding new ways to diagnose disease and to fight cancer. Soon she expects to formalize an affiliation with the David H. Koch Institute for Integrative Cancer Research, which focuses on finding ways to apply quantitative techniques to fighting cancer.
Sikes says she feels at home at MIT, but the road has been long. "My challenge was learning three different disciplines -- physical chemistry, engineering, and enough biology to be able to work in biological systems," she says. "To make a technologically useful product, you don't always need the engineering. But that was important to me as well."
Photo (top): Brittany G
Susan Gaidos writes from near Portland, Maine.