Volume 6, Issue 48 November 2009
The Stories in DNA
Montgomery Slatkin is also a fellow of the Berkeley Center for Theoretical Genomics.
The weave of cloth in a dress, the type of stone in a wall, even the rings in a tree have volumes to tell about their own history. The same holds true for the fabric of life itself—the patterns of nucleotides that make up DNA.
Professor of integrative biology Montgomery Slatkin knows this full well. A mathematician, Slatkin has worked on problems from ecology to paleontology. But as genetic sequence information has become more abundant, Slatkin has turned his attention to tales told by the human genome.
For much of his work, Slatkin predicts how much genetic variation will be found in a population due to various scenarios, and compares those predictions against actual population data. "We try to look at a large number of genes in different individuals to try to figure out what past events could account for what we're seeing," he says. His research offers an uncanny glimpse into human evolution, genetic diseases, and the power of modern forensics.
One question that has long puzzled anthropologists is whether our ancestors interbred with Neanderthals. As part of the Neanderthal Genome Project, Slatkin is analyzing whether the genome sequence of these extinct hominids contains any evidence of mixing with modern humans.
Slatkin is researching whether Neanderthals contributed their DNA to the human genome. Image credit: Luna04, Wikimedia Commons
If so, he and other collaborators on that project should be able to tell approximately when and where that occurred. If the Neanderthal genome turns out to be more similar to that of Europeans, the mixing likely occurred about 30,000-40,000 years ago, when the two coexisted in Europe. If Neanderthals are just as closely related to Asians and Europeans, the mixing likely occurred earlier, in the Middle East, "before the ancestors of Europeans and Asians diverged from one another," Slatkin says.
Slatkin is using similar techniques to speed the hunt for disease genes. Cancer, autism, Alzheimer's, diabetes and heart disease are all thought to be caused by the interactions of multiple genes. The catch: scientists can't seem to find enough associated genes to explain the prevalence of each illness.
For example, says Slatkin, "in the case of type 2 diabetes, fewer than 20 genes are known to be associated with increased risk of the disease. Yet together they only account for three percent of cases."
California and other states use DNA markers like these to find potential crime suspects. Slatkin is studying whether state policies on cold hits are adequately supported by science. Image credit: courtesy PaleWhaleGail, Wikimedia Commons
This missing heritability problem has several possible explanations. One is that scientists haven't searched hard enough for the remaining genes. Though studies might survey the genes of perhaps 20,000 patients, they may need to include 200,000 patients or more to detect the minor effect of each contributing gene. The studies are currently designed to detect disease genes prevalent in 20 to 30 percent of patients. However, just a few percent of patients might carry them. "If so, the methods that we're using just aren't going to find them," Slatkin says.
Slatkin is trying to predict the kinds of observations that would suggest uncommon disease genes are important, and investigating how studies might find such genes. His results could affect how scientists approach genetic disease research.
Genetic information is now so easy to obtain that it is being used by the criminal justice system to identify crime suspects. California, like other states, maintains a database of DNA from convicted felons. Investigators can search the database for DNA fingerprints that match crime scene evidence. Slatkin and colleagues Erin Murphy, of Berkeley Law, and Yun Song, of Statistics and Electrical Engineering / Computer Science, are evaluating how effective the state's written policy is at finding relatives of people whose DNA fingerprints are in the California database. A match for all 13 marker genes is almost certainly the perpetrator. But a partial match can point to another good lead—a brother of the man in the database.
"Now the issue is much less clear," Slatkin says. "At what point is the likelihood that it's a brother so high that law enforcement agencies are entitled to do what is sure to be a very intrusive investigation? If we cannot be entirely sure it was his brother, how sure do we have to be?"
Slatkin and his collaborators found that a familial search can identify a guilty brother 90-95 percent of the time. However, the search also is likely to turn up 5 to 10 unrelated people in a state of California's population size.
"Some will be six-month-old babies, and some will be 95-year-old men who are automatically disqualified. But there is a significant risk of picking up an individual who may have a brother, a man who might not be a very nice person, but who isn't the guy who committed the crime, and that has to be taken into account," Slatkin says.
Though the policy requires the use of 13 markers, the state laboratory that does the comparisons examines up to 15 genes. The extra markers greatly reduce the risk of false positives. However, says Slatkin, as long as the written policy specifies otherwise, "it sets a dangerous precedent for other states to follow." Even the long arm of the law must respect the gene patterns that reveal so much about our populations and pasts.