The number of different bacteria and viruses that can attack the body is practically infinite. Understanding how the body produces immune cells in sufficient variety to ward off this vast array of pathogens -- and how this process can go awry and produce leukemia -- would be a scientific triumph and medical prize.
The problem, however, does not yield to a simple set of experiments because of the huge diversity of germs, as well as of the white blood cells that attack them. That's why Lindsay Cowell and Garnett Kelsoe work together. She is an assistant professor of biostatistics and bioinformatics with expertise in computational methods, and he is professor of immunology with a wet lab that experiments directly with immune cells.
Together, they discovered an important piece of the puzzle of immune cell diversity. Using mathematical models by Cowell and experiments by Kelsoe, they uncovered previously-hidden genomic markers in lymphocytes, a type of white blood cell. The markers play a role in the process by which DNA breaks apart and then recombines in lymphocytes to adapt to new pathogens. By identifying the location and function of these markers, the scientists could more easily differentiate healthy cells from cancerous ones in lymphocytic leukemia.
"I think we wouldn't have been as successful working separately," Cowell said. "If I were doing this in the biostatistics department without working with Garnett, there would be no way to experimentally verify [her mathematical models]."
Kelsoe said that without Cowell's computer-generated predictions to guide his experiments, he would have "30 billion places" to test in the mouse genome.
"Until the dean gives me more money, that's too many," he said.
Such collaboration is not unique at Duke.
Joe Nevins, a professor of molecular genetics and microbiology, and Mike West, a professor of statistics and decision sciences, teamed up to study breast cancer. They developed a guide for doctors to choose among options for treating breast cancer based on how a woman's genetic profile affects her likelihood of tumors recurring.
Another team is Susan Murphy and Terry Furey. Murphy, an assistant research professor of gynecologic oncology, and Furey, an assistant research professor of biostatistics and bioinformatics, are searching for the genetic culprits of a disease that afflicts millions of women: ovarian cancer. By combining her data and his computational expertise, they're identifying genes likely to be turned off abnormally in ways that can lead to ovarian cancer. Furey's computer predictions guide Murphy as she experiments to clarify which genes are involved.
Assistant professor of biomedical engineering Lingchong You put together his own expertise in both biology and engineering to create a system that genetically regulates the reproduction of bacteria. Such a "gene circuit" might help control the spread of malignant cells in the human body or the environment.
The new approach to research even shows up among undergraduate students like senior Jackie Ou. A math and biology double major, Ou works in Professor Fred Dietrich's bioinformatics lab with graduate student Charles Hall, researching how genetic information may be passed among bacteria in ways besides traditional inheritance.
"I thought it was fascinating how you could use both computational and experimental tools, bring them together in one place, to answer one question," Ou says.
Duke's provost and senior academic officer, Peter Lange, sees such an approach playing an important role in the university's academic future.
"Bioinformatics and computational biology offer particularly good examples of how universities must embrace new approaches to interdisciplinary research even as they strengthen their departments," he says.
The new frontier between genetics and mathematics extends beyond individual research laboratories to Duke programs and curricula. In 2003, for example, Duke's Institute for Genome Sciences and Policy launched a doctoral program in computational biology and bioinformatics.
"We're trying to train a whole new kind of person that didn't exist ten years ago in the vast majority of computer science or biology departments," says Terrence Oas, the program's director of graduate studies and an associate professor of biochemistry and chemistry. By enlisting faculty from both the biological and computational sciences, the program is "trying to create new scientists that are hybrids of current scientists," Oas says.
Other bioinformatics courses and seminars continue to sprout up. The scientists creating the classes say dual training is crucial for students to participate in the genomic revolution.
Scott Schmidler teaches a "statistical methods in bioinformatics" course in Duke's Howard Hughes undergraduate program on "Making Meaning of Genomic Information."
"What's happening in the biological sciences in general is that they're becoming quantitative," says Schmidler, an assistant professor of statistics and decision sciences. "It's not sufficient just to be trained in laboratory techniques!. It's also not sufficient the other way around."
Schmidler points out that a number of common steps in genomics research, such as matching sequences of DNA or sifting through a soup of DNA material, involve statistics.
"You do one experiment, you [get] 10,000 data points," he says. "You're looking for one or two really different, real effects out of 10,000 data points. That's an inherently statistical problem."
Computing expertise also is essential for researchers who seek to take advantage of large central databases with genomic information from laboratories around the world.
Because new findings are constantly being added to these central databases and distributed to specialized databases, biologists need to be proficient in using "genome browsers" to know which experiments are worth doing, says Simon Gregory, an assistant research professor at the Duke Center for Human Genetics.
A laptop "for a lot of people should be their first [lab] bench, before they dive in ! and do experiments," says Gregory, who organizes annual workshops on bioinformatics. Similarly, Duke students who major in computer science can now pursue a minor in computational biology.
"The scale of data that people manage is enormous," says Owen Astrachan, a professor of the practice in computer science who teaches a course for first-year students on computer science and programming from a genomics perspective. "How to deal with a terabyte of information is something computer science can teach."