Driven by Genomic Complexity, Old Barriers Between Disciplines Wear Thin.
Multiply the more than 3 billion base pairs in the human genome by the almost 7 billion people living on earth and it is clear that the study of genomics has almost unlimited potential to generate data. As genome technologies continue to bring us ever closer to the $1,000 human genome, those data are getting easier and easier to come by all the time.
And that’s even if you ignore the other hundreds of thousands of species that we share the planet with, each with their own genomes. How to even begin to make any sense of it all is as much a computational question as it is a biological one.
“The data are ballooning not only because of deep sequencing, but also because more people in general are applying genome technologies to their research,” says Uwe Ohler, a faculty member in the Department of Biostatistics & Bioinformatics (B&B) and one of the original computational biologists recruited to the IGSP. “You can’t really do high-throughput genomics experiments without doing computational biology downstream.”
Success in genomics and its various derivative ‘omics increasingly requires an integrative team approach to science, in which experts in genome biology or genomic medicine work equally alongside experts in computational biology, statistics and mathematics. It’s exactly what inspired the creation of the Duke Institute for Genome Sciences & Policy back in 2003. And increasingly, in the genomics world in general and at the IGSP in particular, some of those old distinctions between disciplines are getting harder and harder – and perhaps less worthwhile – to draw.
“Computational biology and genomics are often treated as two separate fields,” says Tim Reddy, who will join the IGSP faculty as an Assistant Professor of Computational Biology in B&B in November. “But I think they work best when they are very closely connected to the point that I try not to even draw a distinction.”
Blurring the Lines
For Reddy, the lines between disciplines blurred long ago. He got his start as a computer scientist. In search of a greater challenge, he found himself increasingly interested in biology, leading him to pursue a doctoral degree in bioinformatics. “I wanted to do something bigger than write programs,” he says. “I got interested in biology simply because it was the most complex system I could find.” So after bioinformatics, he headed for genomics.
His research is focused on understanding how transcription factors turn genes on and off on a genome-wide scale. He is also particularly interested in understanding how hormones, and especially the stress hormone cortisol, influence those genomic controls.
His approach to the problem combines genome sequencing with bioinformatic, biochemical and genetic techniques. On any given day, Reddy might find himself working at the laboratory bench in the morning and programming at the computer in the afternoon.
“It’s a constant interplay on a daily or even hourly basis,” Reddy says. “With the way genomics is evolving, it’s very powerful to be able to do a lot of analysis and experimentation in the same lab.”
Once it is up and running, his IGSP lab will be split about 50-50 between wet and dry bench space. He anticipates computational and experimental students working side by side, with each learning the skills of the other.
Reddy’s scientific style and somewhat roundabout path to a career in biology is in many ways a sign of the times.
“A major impact of the Human Genome Project has been not only to redefine science in terms of digital data and lots of it, but also to change the description of the life scientist of the future,” says IGSP Director Hunt Willard. “He or she is just as likely to come out of a background in math or physics as out of a typical ‘biology’ background.”
A visit to the IGSP, and especially to the labs of some of its newest and youngest faculty members, yields plenty of living proof of Willard’s point.
Take Nicolas Buchler, an Assistant Professor of Biology and Physics in the IGSP. Five years ago, he was strictly a theorist. He never really felt the need to prod bacteria or yeast until he developed an idea about how simple interactions between pairs of proteins might generate all-or-nothing responses. It worked in theory, but he wanted to know if it also worked in the “real world” of a laboratory dish. Rather than waiting for someone else to do the experiments, he decided he had better test it out himself.
“If you understand mathematical science, you can do things in biology that are transformational.”
Now, his Duke lab continues to develop his theories and to perform the hands-on experiments to test them. As in the lab Reddy envisions for himself, some of Buchler’s students spend their days at the lab bench, putting simple circuits into yeast, while others test the same ideas in the “parallel universe” of the computer.
Lingchong You, a chemical and biomedical engineer by training who joined the IGSP and engineering faculty back in 2004, also found himself attracted to the complexity of biology. His research team designs synthetic gene circuits meant to guide cell behavior much as computer software directs a computer.
“In the case of computers, we know exactly how they are constructed,” You has said. “We’ve put the components together and have a good understanding of how a computer behaves. In synthetic biology, we may know a lot, but our understanding is far from complete.” So, to see whether the synthetic programs work as anticipated, he loads the circuits he designs into E. coli bacteria in the lab.
More New Blood
This fall, the IGSP will also welcome Raluca Gordân as another new faculty member in B&B. Gordân’s interest in biology goes back to her days as a computer science undergraduate in her home country of Romania.
“I started doing genetic algorithms,” she says. “It was kind of backwards. I was applying genetics to computation.”
Her interest led her overseas to Duke, where she did her doctoral work under Alex Hartemink, himself one of this new breed of scientist with a background in engineering and computer science, faculty appointments in computer science, statistical science and biology, and affiliations with the IGSP and the Center for Systems Biology.
As a graduate student, Gordân designed and applied computational methods for exploring interactions between proteins and DNA and their influence on gene expression. She has since gone on to a postdoctoral position where she is designing and carrying out laboratory experiments to test her computationally derived hypotheses about the binding relationships between transcription factors and DNA sequences.
“It’s exciting to get my own data – to be the first to see it,” she says. She plans to bring the protein binding microarray technology she has been using at Harvard back to her brand new lab in the IGSP.
It can be argued that the increase in scientists from varied backgrounds to the study of the genome sciences changes not just the makeup of the scientific community, but also the nature and direction of the science itself.
“At first I thought computational biology was just about analyzing biological data to try to answer questions,” Gordân says. “But some questions actually come from the data. Computational biology doesn’t just answer questions; it can help you come up with questions that had never been thought about before.”
“Computational biology doesn’t just answer questions; it can help you come up with questions that had never been thought about before.”
For instance, she has uncovered evidence in the data that transcription factors bind DNA not just based on short sequences known as “core binding motifs,” but also on neighboring sequences.
“You can’t capture it with a simple model,” she says. “It’s not just about where a protein interacts directly. It has more to do with DNA structure and how the DNA needs to bend or change its structure. DNA binding sites are never isolated in the genome. They are always in context, and that seems to be important.”
The idea that information and computation have the power to be transformative is not news to Robert Calderbank, Dean of Natural Sciences and Philip Griffiths Professor of Computer Science.
“Everyone will admit statistics and computation are useful, but what’s important is that they are actually strategic,” Calderbank says. “If you understand mathematical science, you can do things in biology that are transformational – things you couldn’t do without that knowledge. That’s the proposition.“
Breaking the Mold
For Willard, this younger generation of faculty really exemplifies the transformation of biology that is surely taking place in the genome era. “The 30-somethings earned their stripes when sequences and gene expression data were not a dream or a promise, but a reality only a mouse click away,” he says. “They don’t know what traditional biology is; they’ve redefined it.”
This is not to suggest that there will no longer be an important place for all those with deep training in one discipline or another.
“There certainly are far more scientists today with training in both experimental techniques and quantitative analysis,” says Philip Benfey, director of systems biology in the IGSP. “We have been fortunate to recruit some of the very best. This being said, I think that there will be a continued place for scientists with deep training in one of these areas and the ability to communicate productively with researchers in the other. I foresee a continuum of scientists, from those who can ‘do it all’ to those who collaborate most of the time. Neither will become the dominant species.”
As a place where collaborations come easy and teamwork is the norm, the IGSP offers a perfect stomping ground for both the old and the new type of scientist. For one thing, Ohler says, the IGSP’s location, with Computer Science and Statistics nearby and the School of Medicine across the street, is very hard to beat.
“The institutes are places where interesting things happen at the borders of disciplines,” Calderbank says. And that’s exactly what drew Reddy and Gordân to Duke as the place to launch their independent careers.
“The interdisciplinary aspect of the IGSP makes it a good fit for me and for the kind of research I want to do,” Reddy says. “There is an openness that I think is unique, and that’s what I’m most looking forward to.”