Title: Research Associate, Department of Biostatistics; Associate
Director, Bioinformatics Core,
Harvard School of Public Health
Education: PhD, University of Cologne, 2004
Recommended by: Winston Hide
Oliver Hofmann likes that his work falls at the "interface of lots of different technologies." Though trained as a biologist, Hofmann has been involved in computer science and programming for quite some time and sees a great opportunity for someone with dual backgrounds. "I think there is a niche to understand these two very, very distinctive, separate communities: biomedical research, in particular, translational research and computer science," Hofmann says.
With collaborators at the RIKEN Genome Institute in Tokyo, Hofmann is studying melanoma cancer stem cells. The RIKEN researchers developed a tool that can, from next-gen sequencing data, determine where transcription actually starts — "not just the canonical start," Hofmann says. This, he adds, "helps tremendously with narrowing down the promoter search and identifying regulatory elements." So far, Hofmann and his colleagues found that there are "a fair amount of genes that have very, very specific transcription initiation sites that are specific to stem cells," he says.
Another project focuses on understanding which genes influence drug therapy and dosage. "From clinical trials, we have a good understanding of what drug treatment is recommended at each time point," he says. In this project, Hofmann and his colleagues will be using next-gen sequencing to get ultra-high coverage — thousand-fold — of a patient cohort that they will then follow over time. From that, they hope to pick out minority variants that affect what drug doses the patients should receive.
Hofmann says his work and methods have been informed by others, such as David Haussler and Lincoln Stein, who are merging wet lab approaches with computational ones.
One of the challenges of straddling different fields is making sure both sides understand the contribution of the other. Computational biologists, he says, often get lost in the shuffle. "We don't generate the data, we don't do the final validation of the data. We drive the hypothesis generation, the validation," he says. To overcome that issue, Hofmann says more biology students should be exposed to computational approaches during their training.
In the future, he hopes that the role of computational biologists becomes clearer. One way he sees this happening is through integration; computational biologists could be embedded into research groups as specialists, much as postdocs with certain expertise are hired into a group. "The drawback is that those people are usually then quite isolated. You're the only person in the lab who is doing this," he says.
The other possibility, he says, is that computational centers could be set up much like genome centers are now. This way, the computational biologists would stay at the cutting edge of technology. "So, saying, 'OK, here's a new technology, we need to figure out a way to analyze it,'" he says. That way, he adds, computational biologists would be working "alongside new developments in engineering."
Publications of note
In a 2008 Proceedings of the National Academy of Sciences paper, Hofmann and his colleagues surveyed cancer/testis gene expression in normal and cancer expression libraries in silico. These genes, he says, are tightly regulated and are only expressed in germ cell development. In cancer, however, they get switched back on and their antigens are often immunogenic. In this study, they find that the CT genes are highly expressed in the testis and in other tissues as well, and they identified three new CT genes.
And the Nobel goes to...
While you can't apply for the Nobel Prize, Hofmann says if he were to win, he'd like it to be for enabling better diagnostics of infectious diseases.