For computational biologists, finding solid collaborators on the biology side is key to actually applying their programmatic skills. Most systems biology studies entail both computational prediction, or dry lab, and functional validation, or wet lab, work. "It's always good to have very close collaborators and to work on very specific problems on the biology side," Duke University's Uwe Ohler says.
Practicing the art of finding a good collaborator takes some effort. As an instructor in Duke's PhD program in computational biology, Ohler tries to impart to his students the importance of having a clear connection between wet and dry lab. "What we try to do is to have the students realize that they have to not just sit there and code stuff," he says. "You really have to think about a specific biological question and a specific problem area where you actually will apply your software to."
Perfecting your coding skills is often not enough. You have to get the scientific community to understand what you're working on, and the larger context of it. "Often these things are more how you present yourself to other people," he says, rather than what you're actually doing.
A larger challenge lies in continuing to legitimize computational biology as a science, especially in light of the funding crunch and ever-increasing competition to win grants. While biologists still define most of the problems that he helps them solve in collaborations, Ohler points out that computational scientists can lead the way, too. From his mentors, and just the act of going from computer science to systems biology, he learned that it's "always good to have very close collaborators and to work on very specific problems on the biology side." Computational biologists are "often still regarded as the helpers in the background, always playing second fiddle in a sense. Biologists define the problems, and then the computational biologists help them get it done." However, Ohler thinks traditional biologists need to open their eyes. "It's simply a different way how we approach it. We think about it genome-wide, we write software code, we look at everything at once, we look at patterns that emerge from the analysis."