Developing Algorithms for Cancer Genes

Gurinder Atwal
Title: Assistant Professor, Cold Spring Harbor Laboratory
Education: PhD, Cornell University, 2002
Recommended by: Bruce Stillman

Gurinder Singh Atwal — also known as "Mickey" — certainly wasn't the only biology student who was frustrated by the mounds of memorization. As a medical student at the University of Cambridge in the mid-'90s, Atwal discovered his real passion lie in quantitative subjects, and he decided to switch gears. After earning a PhD in theoretical physics, he's now turning that mathematical mindset toward some hard-to-solve biological problems.

Atwal is developing mathematical and computational models to study what governs genetic variations in systems, ranging from single cells to entire populations. For the past couple of years, he says that he's "been focused on looking at genetic variation in people who are at a higher risk of developing certain kinds of cancer." Specifically, he's interested in the variability of genes that modulate the effect of p53. While there's lots of data — and it continues to pile up — he's making headway.

A few years ago, Atwal noticed that "the distribution — the variability — of genetic variants of oncogenes seemed odd, in the sense that the variability [in these genes] were much smaller than expected." He developed a hunch that speculated that some of these genes were under evolutionary pressure. To test his idea, he developed a mathematical model of a set of genetic variants evolving under selection pressure, and asked the questions, "What's the expected variability of the variants in these genes if there was no selection pressure, and what's the expected variability if there was selection pressure?" After coming up with a simulated set of variants, he had hit the nail on the head. "It turns out that my hunch was correct: a lot of these genes did seem to be under positive natural selection." Further studies showed that p53 controlled implantation rates in mice and that variants in the network were associated with infertility in women.

While at Princeton University, Atwal says his postdoc mentor, Arnold Levine, helped him learn how to straddle the disparate realms of computational science and biology. Since Levine is an experimental biologist, "he wasn't able to help with the computational tasks, but he did have a very good intuition, and more importantly, he was able to keep my feet on the ground," Atwal says.

Looking ahead

Still, many computational biologists remain out of the loop. "It's still two cultures: people with background in mathematics, physics, and computer science, and people with background in experimental biology," Atwal says. "There's still a bit of skepticism, at least on the part of biologists, as to what a quantitative person can do. So it's up to us to break down those barriers and learn a common language."

Publications of note

In a paper published last June in the Proceedings of the National Academy of Sciences, Atwal and colleagues studied the effect of mutations in the human oncogene MDM4. Their algorithm detected candidate SNPs, which then linked to increased risk for breast and ovarian cancers in Ashkenazi Jewish and European cohorts, respectively. Then, Atwal says, "we hypothesized that maybe it has a role in reproductive success if it's under [selection] pressure." After genotyping women at an IVF clinic in Manhattan, "it turned out that this allele was quite indicative of how successful women were in conceiving or not," he says. "It was a double whammy, this allele — not only did it confer risk of developing tumors over a lifetime, but it also conferred a risk of implantation failure."

And the Nobel goes to...

If it actually was a group from Sweden on the phone, Atwal hopes to have won the Nobel Prize for "understanding the role of complex evolution in developing complex phenotypes."