Posted by on August 1, 2014 in AncestryDNA

AncestryDNA customers with significant Jewish ancestry have witnessed the challenges that we and other genetic genealogy testing companies have faced when predicting genetic relatives. Most Jewish customers find that we predict them to be related to nearly every other Jewish customer in the database! So while we all know that the cousin matches for Jewish and some Hispanic customers were over-estimates, detecting which cousin matches were real and which ones were bogus has always been a challenge for these populations.

The AncestryDNA science team has been unsatisfied with the cousin matches we have delivered to many of our customers and as part of our continued commitment to bring innovative genomics to you, we are pleased and proud to tell you that we have found the first solution to the “overmatching” experienced by Jewish, Hispanic and other customers.

Screenshot_8_1_14__8_48_AM

When you take a step back, matching isn’t as simple as it might first appear.  After all, we are all 99% identical. In other words, determining which parts of our genome make us “human” and which make us “recent cousins” is tricky and at the heart of the cousin matching issues for customers of Jewish and Hispanic ancestry.

In DNA matching, we are looking for pieces of DNA that appear identical between individuals. But there are a couple of reasons why it could be identical. For genealogy research we’re interested in DNA that’s identical because we’re both descended from a recent common ancestor. We call this identical by descent (IBD).  This is what helps us to make new discoveries in finding new relatives, new ancestors, and collaborating on our research. However, we also find pieces of DNA that are identical for another reason.  At one extreme we find pieces of DNA that are identical because it is essential for human survival.  At the other, we find pieces of DNA that are identical because two people are of the same ethnicity. We call these segments identical by state (IBS) because the piece of DNA is identical for a reason other than a recent common ancestor. This, we have found, often happens in individuals of Jewish descent. Given the historically small population size of the Jewish community, two Jewish individuals might have a lot of DNA that looks to be identical.  But that identical DNA might only be because of their shared ethnic history – in other words, identical by state, not identical by descent.

The challenge in DNA matching is to tease apart which segments are IBD, and which ones are IBS.  How did we do it?  By studying patterns of matches across our more than half a million AncestryDNA customers, we found that in certain places of the genome, thousands of people were being estimated to share DNA with one another.  This isn’t a hallmark of thousands of people actually being closely related to one another.  Instead, it’s likely a hallmark of a common ethnicity.  Our scientific advancements using such insights from more than half a million people have allowed us to effectively “pan for gold” in our matches – by throwing out matches that appear to only be IBS, and keeping those that are IBD.

What does this mean for you? 

While the problem was more pronounced in customers of Jewish and some Hispanic descents, we observed this problem across all ethnic groups.  So, all customers will see increased accuracy of their DNA matches, and significantly fewer “false” matches.

Eager to see your new set of DNA matches?  It will be available in the coming months, and we’re planning to email our existing AncestryDNA customers when the new matching results are ready with more information about what to expect and what it means for your research. So when the time comes, we’re excited to hear about the new family history discoveries you’ve made or distant cousins you connected with through the advancements of our updated matching service. I’m expecting a lot of great stories will surface, and we can’t wait to hear yours.

 

About Ken Chahine

Ken Chahine, Ph.D., J.D., has served as Senior Vice President and General Manager for Ancestry.com DNA, LLC since 2011. Prior to joining Ancestry, he held several positions, including as Chief Executive Officer of Avigen, a biotechnology company in the Department of Human Genetics at the University of Utah, and at Parke-Davis Pharmaceuticals (currently Pfizer). Ken also teaches a course focused on new venture development, intellectual property, and licensing at the University of Utah's College of Law. He earned a Ph.D. in Biochemistry from the University of Michigan, a J.D. from the University of Utah College of Law, and a B.A. in Chemistry from Florida State University.

18 Comments

scwbcm 

Serious scientists and genealogists alike share the intellectual self control to take the time to try and understand something and to continue to work on improving it. Rushed science is frequently embarrassing, creates expensive mistakes and damages the credibility of the researchers. I applaud your willingness to try to improve this.

August 1, 2014 at 11:27 am
Michael Ward 

This is great to hear! I’m certainly looking for any help possible in organizing my AJ sixth cousins into real and not-so-real relatives. They all cluster in the upper parts of chromosomes 5 and 12, but the ones that are -not- from my Polish ancestry are the ones that really make it confusing — some of those ethnic-German Mennonites on my father’s side must have some AJ ancestry as well.

August 1, 2014 at 12:28 pm
Kalani 

I’m curious how this would work really. Of course I noticed this and was questioning exactly how much of these matches are REALLY IBS? People reacted negatively to me giving examples of a given group, sometimes not related ethnic groups.

But as far as endogamous populations, how would that even work? You don’t even provide a chromosome browser and your estimates of these ‘hints” are based on only a similar looking name and a DNA match. I come from a heavily endogamous group even more than AJ and we lack genetic diversity to where we DO know we all come from common ancestorS. How would you handle our situation?

August 1, 2014 at 2:08 pm
Nancy Schlegel 

Thanks Ken & team, for working on this since the release of Ancestry DNA, and coming up with what sounds like a promising solution!

Waiting on pins & needles now, for the email announcing our results have been updated with this change :-)

August 1, 2014 at 2:10 pm
K S Rose 

I’m very happy to hear this! I am assuming French Canadians will also be adjusted?

One comment though, IBD segments are also IBS. It’s important for customers to understand this and that it is how a segment gets to be IBS, whether IBD or not, that matters. I think it’s excellent that you mentioned that it has to be from a recent common ancestor but it would be even better if you called them IBD or non IBD segments.

August 1, 2014 at 8:46 pm
Michael 

This is so awesome! I manage 5 tests and they have a mix of real matches that have bashed down brick walls along with the tons of bogus matches. I would be delighted to be an early tester to help shake things out.

August 1, 2014 at 10:38 pm
John 

Ken,

I’m thrilled to hear of the upcoming improvement you have just announced. I have a total of roughly 9500 potential matches, so anything to increase the accuracy to help find more cousins will be very welcome!

However, the customer base of AncestryDNA is essentially American as I understand it. Which means that the longer one’s ancestors have been in the US, the more cousins would likely be found. But if one’s family has not been the US for very many generations, then that will limit the number of true matches.

In my case, my parents (both Canadian) settled in the US in 1946 and my earliest ancestors to come to this part of the world, came to North America from the UK in 1885. So there are some cousins in the US and Canada; but, though some are known, there must be many more that are unknown in the UK. Especially since the test puts me at 83% Great Britain and 9% Ireland.

So if AncestryDNA were to make the the test available in the UK and Canada, I would be on pins a needles with excitement!

Are you planning to do so, and if so, when?

And since Ancestry.com has branches in Australia and in Europe etc. etc. what about testing in all those other countries?

August 2, 2014 at 1:03 am
Karen 

Without offering chromosome browsing to AncestryDNA clients I don’t see how matching is ever going to improve. Ancestry participants have been struggling with this issue far too long. Those who want to do real matching still must upload to GEDMATCH or transfer to FTDNA. I have folks telling me “I now know I descend from the “ABC” line of my surname because I found a match to someone on AncestryDNA.” If that were to be true, I match every single unrelated line of my mother’s surname. Of course it isn’t true, and even though we share that particular surname, the match is elsewhere. It’s easy to see that in cases where YDNA testing has proven the lines. In a lot of cases t isn’t possible to rule out these kinds of coincidental matches without chromosome browsing.

August 2, 2014 at 6:05 am
Marci 

I second Karen’s comment. Without chromosome matching tools, you DO NOT HAVE A MATCH. Period. You can find Ancestry “matches” all day but until you triangulate them on a chromosome string, they mean nothing. I expect any of us who have taken the time (and spent the additional money) to use GedMatch, FTDNA or 23andMe can attest to how many Ancestry “matches” fall apart under scrutiny.

August 2, 2014 at 10:07 am
Judy 

What if you possibly have Jewish ancestry that ‘is not’ Ashkenazie? Is something in the works to distinguish Sephardic ancestry? Or Mizrahi?

Thank you.

August 3, 2014 at 10:41 pm
Cool GenStuff - Monday 4 August 2014 | Hack Genealogy 

[...] DNA: AncestryDNA Matching Update Impacts Jewish Ancestry (via Ancestry [...]

August 4, 2014 at 8:50 am
Diane Jacobs 

While I appreciate your recognition that your matching is poor, I wish you had figured out how to do matching before you solicited our business and took our money. We have found no value in your DNA testing/matching.

August 4, 2014 at 12:22 pm
Peggy Deras 

Being of partially Mexican descent, with some Ashkenazi flavoring, I welcome your announcement. However, I still believe my Ancestry DNA test is essentially useless without a chromosome browser result to compare with matches from other services in my spreadsheet.

Ancestry? That’s where I go to post the new relationships I find with other DNA services to my tree.

August 4, 2014 at 1:10 pm
mel 

I also wish that you had discovered this before promoting the DNA testing. I have many “matches” but have not found even one that is “real” in any sense. Many of the matches do not even have family trees and a majority of the “third cousins” do not even respond to my e-mails. I have found NO value in this DNA test whatsoever. Very disappointing

August 4, 2014 at 1:20 pm
AncestryDNA Upgrades the Methods Used for Cousin Matches 

[...] Read the full blog post. [...]

August 4, 2014 at 3:29 pm
A Piper 

I agree with the people posting requests for a chromosome browser. Without one, all I know is that “maybe” I match someone that Ancestry has matched me to. From there, I need to convince the match to either upload to Gedmatch or test somewhere else.

August 7, 2014 at 7:46 pm
The DNA matching research and development life cycle 

[...] continues to grow rapidly at an exponential rate.  While that means millions of opportunities for personal discoveries by AncestryDNA members, it also means a lot of data that the AncestryDNA science team can put back [...]

August 19, 2014 at 2:24 pm
Edgar 

I’m sorry I don’t buy the distinction between IBD and IBS. The reason people of same ethnicity share genetic matter, is precisely because they descend from a common ancestor! Ethnicity is the result of common ancestry. Although I have no scientific proof for my assertation, it is the most straightforward explanation for shared ethnicity and the burden of proof to make your distinction and claim there is such a phenomenon as shared DNA due to “the state” is upon you. The United States is comprised of many ethnicities and I don’t see ethnicity changing because we all live in the same “state.” I believe you may counter with the argument that after billions of years changes in ethnicity will occur. Very Darwinian and very unhelpful for genealogy. I would love to see proof for IBS, but until then, I will consider shared ethnicity to be the result of shared ancestry.

September 2, 2014 at 6:00 am