Longevity debate: Chips to blame?

At the heart of a feverish debate over the validity of a linkurl:recent genome-wide association study (GWAS) of centenarians;http://www.the-scientist.com/blog/display/57543/ is the authors' possible misuse of gene chips in different testing groups, part of an ongoing issue affecting other GWAS research. How this variation might impact the validity of the longevity findings, however, including the 150 SNPs associated with extreme longevity, is unclear. SNP chips are at the center of the longevity study's faultsImage: Wikimedia Commons, Magnus Manske The initially heralded study, published in Science Express July 2, examined the genomes of about 1,000 centenarians, those rare humans that have reached the age of 100, and compared them to those of controls. However, the authors used two different gene chips to analyze their centenarian population, with one gathering about 10 percent of the data. The data from the control population was also gathered using a total of four different chips. Each chip has its own special quirks, even if it's made by the same manufacturer, so comparing the results from two different chips can be a challenge. If a chip has an issue with a particular SNP, it could produce faulty results, such as incorrectly significant alleles. If scientists compare those data to results from a chip that does not have a problem with that particular SNP, it may look as if one sample is genetically different from the other, when it's simply a misreading from one chip. The authors of the longevity study -- linkurl:Thomas Perls,;http://www.bu.edu/alzresearch/team/faculty/perls.html the director of the linkurl:New England Centenarian Study,;http://www.bumc.bu.edu/centenarian/ and Boston University biostatistician linkurl:Paola Sebastiani;http://people.bu.edu/sebas/ used Illumina's HumanCNV370 chips to analyze the first 90 percent of centenarians, and the company's Human610-quad chip for the last 10 percent. (After the Human610-quad chip was introduced, Illumina stopped making the Human CNV370.) Scientists such as GWAS experts linkurl:David Goldstein,;http://mgm.duke.edu/faculty/goldstein/ a researcher at Duke University, and DeCODE founder linkurl:Kári Stefánsson;http://www.time.com/time/specials/2007/article/0,28804,1595326_1595329_1616840,00.html claim that the Human610-quad chip has been plagued with these quirks. Consequently, it's possible that some of the results -- namely, that the 150 selected SNPs can predict if a person is a centenarian with 77 percent accuracy -- are not valid, Goldstein said. "When you have an imbalance in the chips that you use between cases and controls, any of the polymorphisms that are genotyped differently between the chips appear to be different between the chips," said Goldstein. However, Perls argues that the results are valid, since he and Sebastiani used both kinds of chips on both the centenarian and control groups. For some SNPs, it's impossible to tell if the chip is accurately reporting the identity of the nucleotide at the SNP correctly, unless the scientists look at the distribution of each of the possible responses at the SNP. A normal distribution would have results distributed among the three possibilities -- two copies of allele one, two copies of allele two, or one copy of each allele -- without the results clustered around one possibility. Some faulty SNPs can look like they are giving a good reading, because the chip is giving a signal that it is detecting something at the location of the SNP, but the reading might be inaccurate, said Goldstein. He explained that sometimes the chip can be 'blind' to one of the two possible alleles, so reads the SNP as containing two copies of the other allele, no matter what nucleotide is actually contained in the sample. "It is now known that that is exactly what happened for the two most associated SNPs" in the longevity study -- namely, 2 SNPs most associated with an increased chance of reaching 100. Those markers, called rs1036819 and rs9576827, are known to give an incorrect reading on the 610-quad chip, said Goldstein. "We have checked on our own data and we see those problems on our own data too." "Everyone who works with [these chips] is complaining," said Barzilai. "On one hand you are trying to wait for this best new technology but on the other hand it could be your biggest nightmare." The company, linkurl:Illumina, Inc.,;http://www.illumina.com/ which makes the chips, has done its own quality control and has information on which of the SNPs give faulty readouts on each of the chips. "Most of our customer communication is through newsletters and our support team," said Daniel Peiffer, senior product manager for Illumina's genotyping applications. "Perhaps from our side this could have been better communicated, but the information was there." Both experts and Illumina agree that if scientists find something interesting, they always need to double-check their results with some other method, like Taqman. "If it looks OK you can't be certain it's OK," said Goldstein. "For really important findings you want to separately genotype it." When asked if he contacted Illumina tech support or verified his data using another technology, Perls said he was "unaware of a problem with these two SNPs (out of 150). And now that we are, we are following up." Even with the two most significant SNPs out of the game, some still have faith in the remaining 148 SNPs that the original study found. "It doesn't mean that what they found was not real," said linkurl:Nir Barzilai,;http://molgen.aecom.yu.edu/index.php?id=45&task=view&Itemid=68&option=com_content who is conducting a very similar study of the genetics of extreme longevity in a group of Ashkenazi Jews. "For the moment, I do not see any serious evidence that all the paper's conclusions are wrong," linkurl:Jan Vijg,;http://www.einstein.yu.edu/home/faculty/profile.asp?id=11318 a gerontological geneticist, said in an email. Perls and Sebastiani are currently working to determine if any of the SNPs need to be re-genotyped. "It is a big mistake to make premature judgments about what the data will end up showing," Perls said. "I'm not going to say what the magnitude of the problem is one way or the other until the issue is looked at rigorously." The reanalysis "should be finished within the week, and it will be passed on to the editors at Science," he noted. "It's a study that took 15 years to get to where it is today and a few more weeks isn't too long to wait," said Perls. "Don't rush it, get the right answer." P. Sebastiani, et al. "Genetic signatures of exceptional longevity in humans," Science Express, July 2010 ** Related stories: ***linkurl:DNA may differ between tissues;http://www.the-scientist.com/blog/display/55827/ [20th July 2009]*linkurl:Selected genotyping platforms;http://www.the-scientist.com/article/display/53259/ [1st June 2007]*linkurl:Whole-Genome Genotyping on Microarrays;http://www.the-scientist.com/article/display/15604/ [4th July 2005]*linkurl:Assays Galore;http://www.the-scientist.com/article/display/14106/ [22nd September 2003]

Longevity debate: Chips to blame?

Interested in reading more?

Become a Member of