Genome-wide Association Studies Must Account for Ancestry

Ask someone where their ancestors were from and odds are that they know. Or maybe not. A new study published in Human Molecular Genetics suggests that scientists using the latest tools to scan the human genome must pay attention to ancestry when analyzing and interpreting their results.

  • Share on TwitterShare on FacebookShare on Google+Share on LinkedInEmail a friend

Feinstein Institute/North Shore LIJ Logo

We believe the ancestral differences should also be examined in the context of any clinical epidemiological study

Manhasset, NY (Vocus) December 16, 2008

Ask someone where their ancestors were from and odds are that they know. Or maybe not. A new study published in Human Molecular Genetics suggests that scientists using the latest tools to scan the human genome must pay attention to ancestry when analyzing and interpreting their results.

Chao Tian and Michael F. Seldin, MD, PhD, of University of California Davis, and Peter K. Gregersen, MD, of The Feinstein Institute for Medical Research, have identified allele frequency differences among ethnic groups and subgroups that can result in statistical errors in genetic studies and alter the outcome of the study.

Dr. Gregersen, head of the Feinstein’s Robert S. Boas Center for Genomics and Human Genetics, said that about half of the Caucasian people in the US who were asked where in Europe their ancestors came from did not have a clue. Dr. Gregersen collaborated with Dr. Seldin to identify a few thousand genetic markers that are powerful enough to point to the region where an individual’s ancestors once resided, a tool that is much more reliable than asking a person about his or her ancestry.

And this finding has more than just genealogical significance. In the Human Molecular Genetics paper, they showed that it is critical to account for ancestry in genome-wide association studies or the results may not be accurate. “We believe the ancestral differences should also be examined in the context of any clinical epidemiological study,” the scientists concluded.

“Until now, scientists have depended on a person’s knowledge of their ancestral roots when they study human populations to look for risk genes,” said Dr. Gregersen. “Even among racial groups there can be significant differences in the frequencies of certain genetic alleles.”

For instance, Dr. Gregersen studies rheumatoid arthritis, and in 2004 his team discovered an important risk gene called PTPN22. In the US, the disease associated variant of PTPN22 is present in about eight-to-10 percent of the population. Since the vast majority of Caucasians living in the US had ancestors who once lived in Europe, the few thousand genetic markers identified by Dr. Seldin were used to track those with European origins. They tested the DNA from hundreds of people in Europe and found that there was an enormous difference in the frequency of the risk variant for PTPN22, according to whether a person’s ancestors are from southern vs. northern Europe. In Southern Italy, for instance, only about three percent of the population had this specific PTPN22 allele. Moving north into Sweden, about 12-to-15 percent of the population carries this variant. This variant increases the risk for rheumatoid arthritis and a number of other autoimmune conditions.

With these same genetic markers in hand – so called “ancestry informative makers,” or “AIMs.” the scientists went after their US sample, asking whether these markers could be used to identify where a person’s ancestors were from – even if the subjects themselves did not know this information. Indeed, the study showed that they could.

Last month, Dr. Gregersen presented findings from their work at the American Society of Human Genetics meetings in Philadelphia. The team showed that these markers distinguish ancestry more precisely than any tool available today. The genetic information is so specific that one could look at genes from an Asian individual living in New York and tell whether their ancestors migrated from Cambodia, Japan, Korea or any other country in Asia.

Related studies have also shown that five-to-10 percent of Caucasians have evidence of mixture with populations of Asian or African descent, and Dr. Seldin’s genetic markers can identify these differences.

Going back to the PTPN22 gene, this team of investigators found that it is not a risk factor for rheumatoid arthritis in the Asian population. That means that if Asians were mixed into a genetic study on rheumatoid arthritis it could greatly skew the findings.

“When you put thousands of these markers together, you get a probability of who comes from Sweden or Southern Italy,” said Dr. Gregersen. “This information is important if anyone wants to use population studies to identify risk genes. These new genetic markers are a powerful tool. The practical consequence is that we no longer have to fully rely on a person’s self report. It will increase a scientist’s confidence that their finding is not a spurious association.”

In addition to the study in Human Molecular Genetics, the investigators published a similar study in PLoS Genetics earlier this year.

Contact:
Jamie Talan, science writer-in-residence
516-562-1232
631-682-8781 (cell)

###