While the advent of Illumina's new 24-sample BeadChip format has meant that consumer genomics companies can double or triple their throughput, some genetic genealogists have voiced concerns about what impact these new arrays might have on ancestry testing.
Mountain, View, Calif.-based 23andMe was the first company to announce that it would move its services to a new, completely custom, 24-sample array in November. The company had previously used an Illumina HumanOmniExpress chip capable of processing eight samples.
This month, both Family Tree DNA and Ancestry.com disclosed that they would also adopt the new format, which Illumina introduced in October. Both companies have used the San Diego vendor's 12-sample HumanOmniExpress chips in their services.
Representatives for Family Tree DNA and Ancestry.com told BioArray News recently that the content on their newer chips would not differ significantly from the previous array designs. But 23andMe's new chip is a different case, as the company's new array contains just over 602,000 variants, more than 300,000 fewer markers than the 967,000 variants contained on its previous generation array.
Catherine Afarian, a 23andMe spokesperson, said at the time that the company introduced the new chip that the firm's custom design would permit the elimination of SNPs that are not informative to its reports or research, and that the fourth-generation chip contains markers that will enable 23andMe researchers to better impute many variants present on larger arrays.
However, that decrease in content has some genetic genealogists concerned, particularly about their ability to use new 23andMe data to connect with relatives via third party analysis tools, as their potential genetic matches might have been tested using other, higher-density arrays.
Both Family Tree DNA and Ancestry.com's next-generation arrays contain about 750,000 variants, roughly in line with the earlier chips employed in their services.
"There is no doubt that decreasing the number of SNPs from almost one million down to just over 600,000 will negatively impact the ability to utilize the data effectively with third party tools, which are extremely important to our community," said CeCe Moore, a professional genetic genealogist who works as a consultant to the programs Finding Your Roots with Henry Louis Gates, Jr. and Genealogy Roadshow.
Moore first raised her concerns in a Dec. 7 post on her blog, Your Genetic Genealogist, where she and Tim Janzen, also a genetic genealogist, compared 23andMe's new array design with previous versions of 23andMe's arrays, as well as the array that is used in Family Tree DNA's Family Finder offering. In that comparison, they noted that 23andMe's new chip design shares less than 310,000 autosomal DNA SNPs with the arrays used in Ancestry.com and Family Tree DNA's services.
Moore told BioArray News this week that this lack of symmetry between 23andMe's new array and the other chips used in similar services, could have a negative impact on 23andMe's sales to the ancestry testing market, currently the only consumer genomics segment it can address, since it suspended all health-related testing while it undergoes a US Food and Drug Administration review.
The FDA sent 23andMe a letter on Nov. 22 stating that 23andMe had not adequately responded to its concerns regarding the validity of its Personal Genome Service, and instructing the consumer genomics firm to discontinue marketing the service until it receives authorization of the agency. On Dec. 5, the company's co-founder and CEO Anne Wojcicki announced on the company's blog that the firm would stop offering health-related genetic tests to new customers, who will only receive ancestry-related results.
Moore noted that genetic genealogists had in the past recommended testing on 23andMe's array, which contained the highest SNP content in the market at the time, and then transfered their data to Family Tree DNA's Family Finder database to connect with other ancestry-testing customers.
"I am confident that Family Tree DNA will do everything in their power to continue to offer this option to their customers," said Moore. "However, if transfers are no longer viable, then this will also hurt 23andMe's sales to these communities since potential customers may choose to initially test at AncestryDNA and transfer into FTDNA's Family Finder database, bypassing 23andMe completely."
Moore also noted that the fact that 23andMe offered the highest SNP content on the market had been an "important selling point" for the firm in the past. "We continually hear feedback in this regard on the community's mailing lists and forums when people cite their reason for choosing 23andMe over the other two main companies in this space," she said, inferring that by transitioning to a lower-content chip, 23andMe could become less attractive to customers who are interested in testing using the highest-density array possible.
"There's definitely going to be some loss," said Tim Janzen, another well-known genetic genealogist, of the compatibility between 23andMe's new chip and those used by other ancestry testing services. "Some matches are going to be dropped, and Family Tree DNA will have to impute that data, or drop those matches altogether," Janzen said.
At the same time, Janzen said that there were a number of advantages for 23andMe in moving to the new array format. In addition to increasing the number of samples it can process on one array, Janzen said that the smaller amount of data generated per sample should allow the firm to reduce its data storage needs.
"Reducing the amount of SNPs should, in theory, mean that data costs should also decrease, because they don't have to manage all of that data," said Janzen. "By having half a million autosomal SNPs instead of 900,000, you cut in half the amount of data you have to manage," he noted.
Moore also said there might be some advantages for customers in 23andMe's new chip, citing the increase of Y and mtDNA SNPs as a potential "boon for genetic genealogists." She also stressed that 23andMe "continues to be an excellent option for genetic genealogists," as "topnotch" features such as the firm's Ancestry Composition feature will not be impacted by the move to the new array.
'The accuracy of our model'
In response to the concerns raised by Moore and Janzen, Afarian told BioArray News that 23andMe has just completed processing for an initial group of customers using the new array and does not yet know if there will be compatibility issues with uploading its data to other services, such as Family Tree DNA's Family Finder.
"Our first priority is to ensure the data from [version 4] provides accurate information to serve as the basis for our ancestry tool and features," Afarian said.
She acknowledged though that the move to the new array will impact identity by descent calculations and "this may impact the number of predicted distant cousins, because in some cases it could be harder to match smaller segments of shared DNA (used to identify very distant relatives)."
Still, Afarian said that such changes are "most likely to be of interest to advanced genetic genealogists with highly technical skills," and said that the "vast majority of our ancestry customers aren't likely to see the changes from version 3 to version 4 impact their overall experience with our service."
Afarian also noted that 23andMe's new chip includes a custom selection of SNPs, meaning that there will be fewer "no calls" in the data. "In addition, we actually are testing more SNPs on the Y chromosome and the mitochondria," said Afarian.
"We are proud of the rigor and accuracy of our model," said Afarian. She noted that the company has provided a white paper on ancestry composition and others on ancestry calculations to make its methodologies transparent, and the company relies on "smart" algorithms, meaning that as more customers use its services, 23andMe's data analysis tools will consider more data, and therefore will become more accurate.
She noted that 23andMe recently updated its Ancestry Composition tool, adding in new reference populations from Africa and Asia, and said that the change in array designs will have no impact on those services.
Overall, Afarian touted the firm's place in the ancestry testing market.
"23andMe was the first genetic ancestry service to provide autosomal analysis," Afarian claimed. "We now have more than half a million customers, offering individuals that many more opportunities to connect with DNA relatives," she said, adding that the firm is "committed to continuing to be an innovator and leader in ancestry genetics."
Janzen, however, disagreed with some of Afarian's comments.
"The issue before us is how much of an impact the reduction in SNPs on [version 4] will impact version 4's compatibility with Family Finder and Ancestry.com data," Janzen said. "It is not only going to be advanced genetic genealogists with highly technical skills who are impacted by this," he said. "It will be every single person who tests at 23andMe on [version] 4 and then is interested in transferring their data to Family Finder or to GEDmatch."
While Janzen acknowledged that the "most aggressive genetic genealogists" are testing with Ancestry.com, Family Tree DNA, and 23andMe, he said that "there are lot of less aggressive genetic genealogists" who choose to test with one or two of the major providers.
Because of the compatibility issue, Janzen said that he is certain that the debut of 23andMe's new array will "at least have some impact" on its sales to genetic genealogy customers. "Exactly how much of an impact this will be remains to be determined," he added.