Detecting complex gene interactions using algorithms
Iwould say predicting who is going to develop cancer is as complex as predicting where a hurricane is going to make landfall when it's sitting out in the middle of the Atlantic Ocean," says Jason Moore, Ph.D. Moore is the founding director of DMS's Computational Genetics Laboratory, as well as the director of the Cancer Center's Bioinformatics Shared Resource.
Predict: Meteorologists at least have developed algorithms that allow them to analyze the complex elements of climate signals—air and ocean temperatures, wind speed and direction, and so on—so they can try to predict where and when hurricanes will hit land.
But only recently have bioinformatics specialists like Moore —who defines bioinformatics as an intersection of biology, computer science, and statistics—begun developing similar mathematical tools. Such tools analyze interactions among genes and between genetic and environmental risk factors for cancer and other common diseases. Since arriving at DMS in August 2004, Moore has been spending half his time developing new computational algorithms and the other half using them to analyze biomedical data.
The good thing about his algorithms "is that they're very powerful and can really detect complex interactions," Moore says. "The downside is that they require a lot of computer time. . . . For a disease like sporadic breast cancer, there are probably dozens if not hundreds of genes that contribute to an individual's susceptibility, in combination with lots of environmental exposures."
To address the computer-time problem, Moore and his team will be developing a 300- to 400- processor supercomputer called a Beowulf Cluster, which entails "stringing together cheap computers to get supercomputer performance at a fraction of the price it would cost to buy a supercomputer."
Statistics: Moore, a national leader in bioinformatics, has advanced degrees from the University of Michigan in human genetics and applied statistics. From 1999 to 2004, he was on the faculty at Vanderbilt, where he was the founding
director of the Bioinformatics and Supercomputing Shared Resource; cofounder and codirector of the Advanced Computing Center for Research and Education; and director of the Bioinformatics Core of the Center for Human Genetics Research.
Under his leadership, DMS's Bioinformatics Shared Resource is developing software and providing computer programming and database support for biomedical researchers. The Computational Genetics Laboratory is developing, evaluating, and applying computational and statistical methods for detecting genetic biomarkers of common human diseases. One project, funded by the National Institutes of Health (NIH), focuses on the interaction of genetic and protein predictors of adverse events following a smallpox vaccination. Another NIH-funded project involves analyzing the genetic basis of trauma recovery to determine why some people develop acute respiratory distress syndrome after a traumatic injury and some don't.
Unwilling: Not all biomedical researchers believe in Moore's approach, however. "You'd be surprised at how many people are unwilling to acknowledge and embrace that complexity," he says. "People are still looking for the silver bullets, the single-gene predictor of Alzheimer's disease, the single gene predictor of breast cancer. For the common diseases, there aren't going to be any silver bullets."
Just as there are no silver bullets for hurricanes.
—Laura Stephenson Carter
If you'd like to offer feedback about this article, we'd welcome getting your comments at DartMed@Dartmouth.edu.
This article may not be reproduced or reposted without permission. To inquire about permission, contact DartMed@Dartmouth.edu.