As a boy, I was always interested in how things work — not so much the natural world, but the things that people made, and how separate parts in an intricate system come together to make the whole function. When I was ten my parents got a computer, one of the original IBM PCs, and I immediately started teaching myself how to program it. For example, I had it play an appropriate tune whenever it was booted up on a holiday.
So computers were really my initial interest, and I decided later to apply those computer skills to biomedical research. Two things got me interested in the biomedical field. First, I had a really great high school chemistry teacher who told me that most everything important had been learned about chemistry and that the future of the field was in applying it to other areas. This included polymer chemistry, such as designing better plastic, and — more interesting to me — organic biochemistry.
The other factor was that when I was in high school, a friend of the family who had helped me learn about computer programming suddenly contracted bacterial meningitis and died within a few days. That really struck me, the fact that these little bacteria could kill a grown man in such a short time. How could that happen, and what could we do to prevent that from happening again? That helped put me on the path to work in biology, to understand how intricate the systems are and how it all works.
When I attended college at Harvard, beginning in 1994, I focused at first on experimental biology but didn't feel passionate about it. I majored in biochemistry and continued to do a great deal of computer work on the side. Around this time, a number of articles were published saying that bioinformatics — using computers to analyze biological data — was the wave of the future. It became clear that this was the path for me, to apply computers to biology.
In 1998, I moved on to Washington University in St. Louis, where I did my PhD work in molecular genetics. The research I conducted there involved taking DNA sequence data of small microorganisms that live in really hot temperatures and trying to make sense of them. On one level it was very academic and impractical, but it taught me important skills about analyzing genetic information.
A New Way of Looking for Genes
During this period, I was becoming more interested in human genetics and understanding how differences in a DNA sequence are expressed in people. When I finished my graduate work, I did postdoctoral training at The Rockefeller University in Jurg Ott's lab, which focuses on statistical genetics. It was there that I started conducting genome-wide association (GWA) studies, which is a new way of looking for genes associated with disease.
The way people have classically looked for genes underlying human diseases is to look within families, study differences in DNA, and try to determine, as the disease passes from generation to generation, how this correlates with who has the disease. For rare diseases that follow traditional Mendelian rules, this process works well. But as early as 1996, people were saying that it did not work well for common, complex diseases, such as cancer, diabetes, and mental illness, because multiple genes are involved, each having a modest affect on risk.
Instead, researchers proposed taking a genome-wide approach, which involves looking for small differences in genetic markers spread throughout the entire genome, comparing the markers present in people with the disease against those who are unaffected. The markers are differences in a single genetic base pair called a single nucleotide polymorphism (SNP). This kind of study, called a genome-wide association (GWA) study, actually became possible only recently, due to advances in computer and microarray technology that allows us to rapidly measure a large number of SNPs in a single person on one single microarray, or chip.
“ An important advantage of being a researcher at Memorial Sloan-Kettering Cancer Center is the opportunity to interact with the clinical side. “
Robert Klein, Computational Biologist
The conventional way of searching for such genes had limited the focus to several hundred pieces of DNA for each person. With a GWA study, you are studying an entire population rather than a family, and you are looking at anywhere from 100,000 to 1 million pieces of DNA. It's a completely different scale in terms of the amount of data that is generated. What's also interesting about the technique is that it can be hypothesis free, that is, you don't say, “I think these genes are the ones responsible for disease risk, so let's look for changes in them.” You can go into a GWA study and simply let the biology tell you what's important.
At Rockefeller, we made the first identification of a disease gene using the GWA approach. The gene, CFH, is linked to age-related macular degeneration, the leading cause of blindness in the developed world. When we published our results in Science in 2005, it really validated the GWA method because there had been skepticism over whether it would work or be worth the cost. After that, the floodgates opened and GWA studies in the few years since then have found genetic markers for a whole spectrum of diseases.
Disease Risk and Cause
In 2006, I came to the Sloan-Kettering Institute to start my own lab in the Cancer Biology and Genetics Program. The main focus of our lab is trying to understand inherited susceptibility to cancer. By identifying the genetic changes that influence cancer susceptibility, we hope to get insight into the early stages of tumor formation.
This would confer two important benefits: First, it would allow us to predict who is at risk for getting cancer, which could guide clinical decisions to reduce that risk. A second benefit, which I find more interesting, is that identifying new genes will suggest new biological pathways that can serve as new therapeutic targets.
We're primarily using GWA studies to see what turns up and then doing follow-up studies to confirm the findings. Although, as I described earlier, you can do GWA studies using an agnostic approach and letting the data tell you what's important, we also don't want to throw out decades of biological research. Instead, we're trying to use what has been learned about the biology of cancer to increase the power of the GWA study.
We're weighting the study based on our suspicion that a gene, or a genetic region, or even a particular SNP, is involved in the disease. For instance, a mutation that we know changes a protein and alters the protein function is much more likely to impact the cell than a mutation that falls in a region of the genome that has no known function.
What Lies Ahead
Although my lab currently focuses on inherited mutations, down the road I'd like to do research on somatic mutations — the changes in DNA that occur as cells replicate during a patient's lifetime. I predict that research findings will indicate that genes associated with inherited cancer have somatic mutations as well that tip a normal cell over the line to become a cancerous cell.
In the next few years, I think researchers are going to start correlating the various SNPs they identify with somatic changes. In other words, molecular subtypes of cancer are going to correlate with different SNPs. So in theory, by looking at a specific SNP, you would be able to tell which type of cancer a person is especially at risk for.
Another development that I think is going to change not just cancer genetics but all genetics is that improvements in sequencing technology will allow us to sequence the entire genome of a person for a relatively low price. Once we can do that, we won't be limited to looking at common variants; we can look at rare variants as well.
This may not have much impact in the short term, but in the long run it could guide treatment for many types of cancer. For example, you would say, “This is the set of 20 mutations we should pay attention to for this type of cancer.” Then by sequencing a tumor cell from a patient with the cancer, you should know which treatment would work best.
An important advantage of being a researcher at Memorial Sloan-Kettering Cancer Center is the opportunity to interact with the clinical side of things. To do the research I do, you need DNA samples from a number of patients, so I collaborate with many clinicians who study a variety of cancers.
Some of my main collaborations right now are with Ross Levine on myeloproliferative disorders, William Pao on lung cancer, Robert Kurtz and Sara Olson on pancreatic cancer, and Manish Shah on gastric cancer. Some of them have patient registries that collect samples from those who fit criteria that make them more likely to have an inherited mutation.
As a non-clinician, it would otherwise take a long time for me to gather so many samples, so it really synergizes nicely to be able to work with people who already have these cohorts.
