Posted by Julie Granka on April 2, 2015 in DNA, Science

At AncestryDNA, we empower our customers to uncover exciting details about their family stories.  Today, we announced a new AncestryDNA experience based on years of research and development by the AncestryDNA science team that is revolutionizing the way people discover, preserve, and share their family history. Learn more about the announcement here.

We are combining DNA testing with the power of 65 million family trees to make it faster and easier than ever to discover ancestors you never knew you had. New Ancestor Discoveries, our new patent-pending capability available with every AncestryDNA test, are a new way to discover your story – finding possible ancestors or relatives for you, even if you know nothing about your family history.

New Ancestor Discoveries provide you with a list of individuals who might be your ancestors – allowing you to build an otherwise empty family tree.
New Ancestor Discoveries provide you with a list of individuals who might be your ancestors – allowing you to build an otherwise empty family tree.

The building blocks: DNA matching and member trees

New Ancestor Discoveries build on two features that are already part of AncestryDNA.  The first building block of New Ancestor Discoveries is DNA matching: the identification of pairs of people who seem to share identical pieces of DNA with one another, and so are likely to be related.

The second building block? While a person’s New Ancestor Discoveries are not based upon any information in your own tree, they are based upon the trees of your DNA matches, or genetic relatives.

Photo two

New Ancestor Discoveries extend from an existing AncestryDNA feature known as DNA Circles™, which integrate these two building blocks.

A DNA Circle is a group of people who all claim to be descended from a particular ancestor (say, William Ogden) in their family trees.  But in addition, they all share DNA with other members in the group.  These two pieces of evidence, and some statistics behind the scenes, support that the members of the DNA Circle really are descendants of William Ogden.

DNAcircle-GRAPH only
Network representation of a DNA Circle. Each icon is an individual in the DNA Circle; an orange line between two individuals means that two people share DNA that they likely inherited from the common ancestor of the DNA Circle.

New Ancestor Discoveries: the power of DNA matching

But what about a descendant of William Ogden who doesn’t know it, and consequently isn’t in the William Ogden DNA Circle?  That is where New Ancestor Discoveries can help – by drawing his or her missing link to the DNA Circle and to William Ogden.

The key idea is that if your DNA matching shows that you are related to a significant number of an ancestor’s descendants in a DNA Circle, you too are likely a descendant – and so should receive a New Ancestor Discovery to this ancestor.

Kathleen Smith would receive a New Ancestor Discovery to this DNA Circle because she matches a large number of people in the Circle.
Kathleen Smith would receive a New Ancestor Discovery to this DNA Circle because she matches a large number of people in the Circle.

For every AncestryDNA member, we examine their DNA matching results to every DNA Circle in the database.  If it looks like they belong to a DNA Circle, they’ll get a New Ancestor Discovery.

That means that the AncestryDNA member who didn’t know she was a descendant of William Ogden might receive a New Ancestor Discovery, and jump-start her family tree.

Photo 5

The nitty gritty

The reality is that if you share DNA with members of a DNA Circle, it does not necessarily mean that you also share the DNA Circle ancestor.  You could instead have another ancestor in common with the Circle members – for example, if the Circle ancestor is the sister of your great-grandmother.  You could also share several different common ancestors with multiple members of the Circle – even if none of them are actually the ancestor of the Circle.

To provide New Ancestor Discoveries that more often suggest direct-line ancestors, we combined several pieces of information into our algorithm: the number of people in the DNA Circle with whom you share DNA, the amount of DNA shared with each DNA Circle member, the number of generations back to the ancestor for each individual in the Circle, and our confidence that you and each member of the Circle share only one common ancestor.

Then, to assess the algorithm’s performance, the science team did what we love and conducted a lot of experiments. For thousands of individuals from a variety of ethnic backgrounds with deep and full family trees, we took them out of their DNA Circles and calculated their New Ancestor Discoveries.

We found that New Ancestor Discoveries correctly identified ancestors in those individuals’ trees, or was related to the ancestor in another way, about 70% of the time. Keeping in mind that as with all statistics, there’s a tradeoff between how many discoveries we can provide to customers and how often they’re correct, this is an impressive feat: we are providing almost 1 million New Ancestor Discoveries to AncestryDNA customers. Even when New Ancestor Discoveries do not point to a direct-line ancestor, they provide a useful starting point for genealogical research.

While someone’s experience with New Ancestor Discoveries may differ depending on their family’s history and how much they already know about their family tree, our testing has convinced us that New Ancestor Discoveries are the way forward: providing a glimpse of your ancestors, or a head start in finding them, as has never before been possible.

As AncestryDNA members update their trees and as the AncestryDNA database grows, we will continue to identify new DNA matches and discover additional and larger DNA Circles.  So the most exciting part about New Ancestor Discoveries is that their impact can only increase over time – bringing the power of DNA matching integrated with family trees to even more AncestryDNA members with diverse and complex family histories.

 

Julie Granka

Julie has been a population geneticist at AncestryDNA since May 2013. Before that, Julie received her Ph.D. in Biology and M.S. in Statistics from Stanford University, where she studied genetic data from human populations and developed computational tools to answer questions about population history and evolution. She also spent time collecting and studying DNA using spit-collection tubes like the ones in an AncestryDNA kit. Julie likes to spend her non-computer time enjoying the outdoors – hiking, biking, running, swimming, camping, and picnicking. But if she’s inside, she’s baking, drawing, and painting.

19 Comments

  1. DNA matching is a powerful tool for finding unexpected connections or proving a relationship. We’ve been asking Ancestry for DNA match data since the beginning, but it’s always been kept secret — except for those who want to put the raw data elsewhere. Does this mean that Ancestry will soon start giving us direct access to our match data? That would be great.

  2. The “Nitty Gritty” section should not be down in the fine print but rather front and center, the first thing a customer reads.

    So far looking at the new “ancestors” created for the tests I manage, none appear to be actual ancestors, but instead cousins, and spouses of those cousins.

  3. Michael

    The concept sounds great! But 70% doesn’t sound too good – does that mean 1 million new ancestor discoveries and 430,000 bogus ancestor notifications? It sounds like the cutoff point could use tweaking for fewer false positives. I’m guessing it’s probably even worse for endogamous populations like mine.

    Recent changes at Ancestry.DNA have started emphasizing quality of matches over quantity, which has been immensely helpful for my research. Please don’t backslide now. Thanks for considering this.

  4. Kelly Wheaton

    I hope that ANCESTRY will rethink this beta offering in so many ways. I believe it may be well intentioned (as were the beat offerings on the Ethnicty Estimate and use of Megabase pairs rather than the industry standard centimorgans). In those cases ANCESTRY showed the wisdom to improve their initial offering. I hope that this will fare likewise. I really had high hopes for this but it is evident that this has many, many flaws. It should be named potential relatives not ancestors, the hyoe should be throttled back and there needs to be an option to delete the proposed ancestir as we get in a Tree hint (no this is not a match). The lakc of true trianagulation on segments or misattribution of a match that is much further back is going to resukt in so much confusion fir Newbies and I fear huge problems with adoptees desperate to find connections. Honestly ANCESTRY had a real opportunity here and by ignoring the serious requests of the Genetic Genealogy community has rolled a better mousetrap that is in no way better.

  5. Larry Van Horn

    I totally agree with the two previous commenters. My biggest complaint so far is how this is presented.to the average Ancestry customer. From your announcement banner presented to me in the test I manage (4).

    We found you new ancestors—just by looking at your DNA. Our latest scientific innovations make it possible to discover ancestors you never knew you had–just through your DNA. It’s an entirely new path to finding your family story.

    In each and every case I can conclusively prove that none of my “new ancestors” you presented to me are in fact ancestors. In fact there is zero chance they could even fit in any of my testers ancestral trees which include my parents and a first cousin, and my spouse and her first cousin. I have two Ancestry “ancestor only” trees involved with these six tests and they are both very mature and well documented. The ancestral lines have been well proven using paper trails and various DNA testing techniques (autosomal/Y-DNA at Ancestry and FTDNA). All my results have been uploaded to GEDMatch/FTDNA and I use those sites to verify and supplement the Ancestry / FTDNA testing results. Bottom line I couldn’t even shoe horn your “new ancestor discoveries” into these two trees even if I had a crowbar or dynamite. As a subscriber to your service for the last 15 years, (I should add that I am also a genealogy instructor and syndicated newspaper columnist), I have seen most of the major changes in the Ancestry product and brand rollout over the years. This new feature you rollout yesterday, was one of the worse I have ever see you do and a major disappointment. I have already wasted a lot of research time on your “new feature,” I won’t be wasting anymore.

  6. Jack Wyatt

    Recently I have discovered a huge human trafficking ring in North Carolina during the 18th Century. It was focused on breeding females to be sold as wives. A limited number of white males were responsible for the breeding while Native Americans, persons of color and probably some unfortunate white persons were kidnapped and illegally enslaved. I am connected to both sides of the fence so to speak, giving me the ability to see what was going on. I would say that the majority of Americans who have ties to the USA or American Colonies from the 19th Century on back have common ancestry with this ring, but they have no way of figuring it out. This likely has a lot to do with DNA circles turning up ancestors that people question.

    • Julie Granka

      Hi Jack, very interesting story. We’ll be interested to see how New Ancestor Discoveries and DNA Circles pan out for these individuals.

  7. Martha

    This new feature is very exciting to me. I do not know, or have any idea, who my father was, but thanks to DNA testing (testing my mother and mother’s side who are very Native American with little to no European blood) I discovered that he must have been Scot-Irish.

    I’m seeing a lot of complaints in the message boards about it predicting impossible connections or connections by marriage. I was able to trace the “New Ancestor” all the way back to Robert I, King of Scotland (and by “I” I mean my daughter who is an attorney and has a history degree from Yale and is not new to genealogical research–she only builds connections when she can point to evidence–the vague nature of this new feature is driving her crazy because she would also like to believe that our roots go back to Robert I, but all she was able to prove was that the “New Ancestor” is a direct descendant).

    For someone like me, who has never known anything about where she comes from, this was very exciting, but now reading all the complaints of impossible connections I feel a little like an orphan whose been told she was getting adopted and then told that it might not happen. I’d like to impress upon you that predicting a connection like this, to a lot of people, is serious, serious business. You’re dealing with lives here, people’s roots, history–not just DNA and family trees (that might be riddled with mistakes). I hope you are taking this new feature very seriously and understand the great impact and potential for devastation that this site is taking on.

    • Julie Granka

      Hi Martha, we’re very excited that you were able to validate the connection between yourself and your New Ancestor Discovery as a direct-line descendant – that’s fantastic. You’ve taken the path that we hope most individuals take when receiving New Ancestor Discoveries – doing the research to determine how this individual could fit into your family tree.

      We take New Ancestor Discoveries, as with all of our products at AncestryDNA, very seriously. We do want to emphasize that while some New Ancestor Discoveries lead to a direct ancestor, some suggested ancestors end up belonging in your family tree as a collateral line relative, and some won’t be closely related to you at all, they likely lived at the same time and place as your actual ancestors – so all three could still be a helpful clue to point you in the right direction. In addition, we find that people who have stronger family tree connections (and so generally have more connections to DNA Circles) will see a much larger proportion of collateral line relatives or suggested ancestors that don’t clearly fit in their trees. But for an individual like you with a smaller tree, a larger proportion of their New Ancestor Discoveries may be to direct-line or other relatives.

  8. Deb

    My dna summary page hasn’t loaded properly since the new ancestor feature was introduced. I was looking forward to seeing what everyone has been talking about. Frustrating!

  9. Marci Bowman

    I understand in another, more recent, statement from the Ancestry team, they admitted that the so-called ancestor matches were based more on matching trees than on DNA evidence. I can find all sorts of amazing things if I believe other people’s trees, but I do my own research instead. For that your card catalog has been most useful and I thank you.

    You could indeed find actual useful matches if you created a spreadsheet for each customer and triangulated their matching chromosome strings with other cousin matches and then compared their trees. That’s pretty much what I do, and it works. But you need a Chromosome Browser to create the spreadsheet and chromosome strings. Now rather than go to all that trouble for so many customers, why not just create the Chromosome Browser and let me see where I match each cousin and I’ll do the work.?

  10. Cheryl Stanfield

    “The reality is that if you share DNA with members of a DNA Circle, it does not necessarily mean that you also share the DNA Circle ancestor. You could instead have another ancestor in common with the Circle members – for example, if the Circle ancestor is the sister of your great-grandmother. You could also share several different common ancestors with multiple members of the Circle – even if none of them are actually the ancestor of the Circle. To provide New Ancestor Discoveries that more often suggest DIRECT- LINE ANCESTORS, we combined several pieces of information into our algorithm….We found that New Ancestor Discoveries correctly identified ancestors in those individuals’ trees, OR WAS RELATED TO THE ANCESTOR IN ANOTHER WAY, about 70% of the time.”

    In order to present how well your algorithm performed, you need to state the percentage of the time DIRECT-LINE ANCESTORS were correctly predicted. I also note there is no indication that source documentation of associated family trees was a factor in selecting the thousands of individuals for this “experiment”. Lots of “full and deep” trees out there that have little basis on fact.

    • Julie Granka

      Hi Michael, Larry, Kelly, Marci and Cheryl:

      Thank you for your comments and feedback. We’ve noticed some confusion around exactly what a New Ancestor Discovery is and is not. To help clarify, it is a suggestion that points you to a potential new ancestor or new relative—someone that may not be in your family tree previously. It is not proof, or a guarantee of a new ancestor. They represent guidance and a starting point for further research.

      While sometimes New Ancestor Discoveries lead to a direct ancestor, some suggested ancestors end up belonging in your family tree as a collateral line relative, and some won’t be closely related to you at all. But in all cases, they likely lived at the same time and place as your actual ancestors so they could be a helpful clue to point you in the right direction.

      In addition, the ratio of new ancestor vs. collateral line relatives in your New Ancestor Discoveries can vary based on how many DNA Circles you are already connected to through your family tree. We can’t provide a New Ancestor Discovery to a DNA Circle that you already in. So we find that people who have stronger family tree connections (and so generally have more connections to DNA Circles) will see a much larger proportion of collateral line relatives or suggested ancestors that don’t clearly fit in your tree, but lived at the same time and place as your actual ancestors. Kenny Freestone expanded on this topic in this post: https://blogs.ancestry.com/ancestry/2015/04/09/new-ancestor-discoveries-clues-not-proof-to-your-past

      Finally, our research has shown that the use of triangulation can be extremely limited. Small segments of shared DNA do not necessarily prove that you are genealogically related to someone, and in addition, you and your distant cousins are unlikely to share the same chromosomal segments inherited from a common ancestor. Please see our new help content on the website that explains this reasoning in great detail. The goal of DNA Circles is to compensate for these limitations to make it possible to consistently group a potentially large number of descendants around an ancestor. Again, keep in mind again that New Ancestor Discoveries and DNA Circles should be considered as additional evidence, and not incontrovertible proof.

  11. Susan Hollis

    Can anyone tell me why my “new” Ancestor has disappeared? He just came to me a few days ago..his name was William Garrett and I had the little circles, etc.but now he is gone? Anyone know why? Thanks, Susan Hollis

    • Julie Granka

      Hi Susan. As with DNA Circles, New Ancestor Discoveries are frequently updated as AncestryDNA members change their trees and as new members are added to the AncestryDNA database. For example, if someone in a DNA Circle makes their tree private or changes the birthdate or other data of an ancestor, in some cases, it’s possible for those changes to affect the existence of that DNA Circle or change its members. So, you may have lost that New Ancestor Discovery because you no longer strongly connect into that DNA Circle, or because the Circle no longer exists. Also, as more people take the test, new DNA matches are found, which could also modify existing DNA Circles as well as their New Ancestor Discoveries. So, keep checking back for updates.

  12. Cheryl Stanfield

    Ancestry’s credibility is suspect because of the continued refusal to provide any scientific support for the matches that are generated. Minimally each match on our lists should include a description of the significant segment(s) of the specific chromosome(s) that have been found in common, along with a help link to a discussion of ancestry’s criteria for supporting the confidence of the match. As it stands, ancestry’s customers are forced to accept what comes out of the “black box” without any scientific support.

    “Finally, our research has shown that the use of triangulation can be extremely limited. Small segments of shared DNA do not necessarily prove that you are genealogically related to someone, and in addition, you and your distant cousins are unlikely to share the same chromosomal segments inherited from a common ancestor.”

    I can agree that small segments of shared DNA may not prove that you are genetically related to someone. I can also agree that it is unlikely that all descendants of a common ancestor will share a common chromosome segment. However, it is very likely that some of those descendants will share a common segment. I have found this to be true working with known cousin groups (documented by paper trail, 2nd to 4th cousin once removed range) using chromosome comparison tools available on both GEDmatch and FTDNA. As one example, several cousins who are descendants of one child of the common ancestor pair share a common segment with several cousins who are descendants of another child of the common ancestor pair. This suggests to me that triangulation can be extremely valuable for determining ancestors when working with extended known (paper trail) cousin groups, and those believed to be cousins based on circumstance.

    “The goal of DNA Circles is to compensate for these limitations to make it possible to consistently group a potentially large number of descendants around an ancestor.” A book could be written on what is wrong with this theory, so here is just one example. What if the common ancestor of that group is not who they think it is? In my efforts to get through “brick walls” I have researched extended families of my ancestors as well as members of the communities where they lived. By tracking down original records (available on line on sites other than ancestry.com as well as visiting archives, etc), I have found over and over again that commonly believed family connections are flat out wrong. I’m talking about connections that have been stated in published genealogies (by local history researchers as well as descendants) that turn out to be based on assumptions (name matching), record abstracts that turn out to have errors, records that were missed by the researchers, etc. These wrong connections have since been reproduced over and over in family trees, so that every family tree out there has these same wrong connections. By discovering these wrong connections over the past couple years, along with uncovering the correct connections, I have gotten through several brick walls.

    There is no doubt that DNA can be a valuable tool for finding your ancestors when used in conjunction with research. Ancestry’s claim for their latest DNA product innovation, “finding possible ancestors or relatives for you, even if you know nothing about your family history” tells me they don’t really care what their current customers think. They already have our family trees and the money we paid for our DNA testing. This new innovation appears targeted towards a new group of customers, probably those who are adopted and seeking their birth parents, and those who aren’t really interested in digging through records, they just want to know if they might be related to anyone famous. What a disappointment this all is, to say the least.

  13. Martha

    Hi Julie Granka,

    Actually, what I said in my previous message was that all my daughter was able to prove was that the “New Ancestor” was a direct descendant of Robert I, King of Scotland–not that we were directly descended from the “New Ancestor.” Confirming this connection will be impossible without more “Ancestors” because as I said in my previous message I have no idea who my father was (zero clues). From our best estimation there are four generations (my father, grandparents, great-grandparents, 2x great-grandparents) between me and this “New Ancestor.” I’ve contacted the DNA circle of this NA (I match 2 known descendants of hers), but there’s really no way to fill in these gaps (if I’m missing something, tips are welcome) without more connections since I have no father from which to build the tree.

    I am happy to hear that the fact that my tree is smaller to non-existent on this side, that it is more likely that the NA is actually a direct-line ancestor. Any and all tips are most welcome.

    Thanks,
    Martha (Previous Message: April 5, 2015)

  14. Jack Wyatt

    Hi Julie,

    In regards to the human trafficking ring selling brides in 18th Century NC that I previously spoke of, New Ancestor Discoveries and DNA Circle users will not be able to figure it out. It will just be another case to them of getting an ancestor that does not make sense. I tested with another service so I am not included in any of this.

    I would like to discuss the human trafficking ring with you. If you wish to, please send a message through the Ancestry.com Message Center to my member ID, 1_cjwyatt.

    Jack

Comments are closed.