Wednesday 12 November 2014

AncestryDNA at Back To Our Past

As mentioned in my previous blog post, I made a special effort at the Back to Our Past show in Dublin to attend the presentation by Mike Mulligan, International Product Manager of Ancestry.com, on "AncestryDNA - DNA testing for family history".

Before I discuss the presentation I just want to start off by providing some background on Ancestry's autosomal product and their entry into the marketplace which you can skip if you're already familiar with the situation in the US. Autosomal DNA tests look at thousands of markers from across your entire genome and can be used to find matches with genetic cousins on all your family lines. These tests are most effective at making connections within the last five or six generations. AncestryDNA is one of three companies which currently offer such a test. 23andMe introduced a cousin-finding feature to their autosomal DNA test in the autumn of 2009. The 23andMe test can be purchased online in fifty-six countries including the UK and Ireland. Family Tree DNA's Family Finder test has been on sale worldwide since February 2010. AncestryDNA began to roll out their autosomal DNA test in the US in the autumn of 2011. They kickstarted their autosomal DNA database by offering free tests to over 10,000 selected subscribers. The test was officially launched in the US on 3 May 2012. I filled out the form to register my interest and received an invitation to order a test in June at the special introductory price of $99. Although the test was at that time supposedly restricted to the US, I was still able to place an order from the UK. This loophole has since been plugged, though you can still circumvent the restriction by using a package forwarder. Tim Janzen has compiled a very useful autosomal DNA testing comparison chart for the ISOGG Wiki which provides further details of the different tests.

The AncestryDNA database has grown very rapidly in the last two years and they have now tested over 500,000 people. Ken Chahine, Senior Vice President and General Manager of AncestryDNA, revealed in a recent presentation that the company are currently selling an astonishing 30,000 to 50,000 DNA kits every month. If the growth continues at the existing rate, they are expecting the database to grow to one million probably by the middle of 2015. This is all well and good if you are American, but a large all-American database is unfortunately not much help for the rest of us unless we are trying to reconnect with the descendants of family members who emigrated to the US in the last 200 years or so. Autosomal tests do also give us matches with more distant cousins, and in fact most of our matches are with people who are predicted to be fifth or more distant cousins simply because we have so many of them. However, finding the genealogical connection with these distant cousins can be difficult if not impossible. While some genealogists are lucky enough to be able to identify all of their 32 great-great-great-grandparents there are probably very few people who can name all of their 64 great-great-great-great-grandparents. I currently have over 7400 matches at Ancestry DNA. Twenty of these matches are with predicted fourth to sixth cousins and these matches have been assigned a confidence level of 95%, which means that there is a very low chance that these matches will be false positives. However, none of my fourth to sixth cousins at AncestryDNA have surnames in common with me and their trees are all in America whereas all my ancestry is in the British Isles so it is an impossible task trying to establish how we are related. My remaining matches are all predicted to be fifth to eight cousins who have been assigned as moderate confidence or low confidence. AncestryDNA add the following note of caution about these distant matches: "Even though there is a 50% (or less) chance that you are related, these matches are still good leads." Nevertheless it seems to me that it would be a futile exercise trying to work my way through all these 7000+ matches with little hope of ever identifying the genealogical relationship, if any. Ancestry do a have a potentially useful "shaky leaf" feature which could help in this situation. If you and your matches have both uploaded trees Ancestry will search the trees for you and will identify the ancestral couples who appear in both trees who might have contributed the shared segment of DNA. However, I do not yet have the benefit of any of these shaky leaf hints.

People are typically getting many more matches at AncestryDNA than at 23andMe and FTDNA, even when you take the comparative sizes of the databases into account. The reason for the large number of matches is that AncestryDNA have set a much lower matching threshold. Ancestry recently announced plans to introduce a new improved matching algorithm, and it is expected that this feature will be rolled out some time before the end of the year. As a result, we can expect our match lists to be drastically pruned but it will be a change that is very much for the better. Blaine Bettinger, who writes The Genetic Genealogist blog, attended a bloggers' summit hosted by AncestryDNA in San Francisco in October where the attendees were given much more detailed information about what to expect. Blaine has provided an excellent write-up in his blog post "Finding genetic cousins - separating fact from fiction". I shall be interested to see how this new feature works out.

We were told at Back To Our Past that the AncestryDNA autosomal test will be launched in the UK and Ireland some time next year. The date has not yet been revealed but my money is on a launch in April 2015 at Who Do You Think You Are? Live in Birmingham. Although the AncestryDNA test is lacking many of the essential tools that we take for granted at Family Tree DNA and 23andMe (for example, a chromosome browser and the ability to download your segment data), Ancestry do have the advantage of a large subscriber base of over 250,000 people in the UK and Ireland. Many of these people will no doubt be tempted to take the test and a lot of them would probably have never considered testing at FTDNA or 23andMe. Family Tree DNA now offer a free autosomal transfer programme though it's necessary to pay a small fee of $39 to unlock additional matches and features. Anyone who tests at AncestryDNA will be able to transfer their results to FTDNA so that they can take advantage of all the extra tools and receive matches in FTDNA's international database. Another option for people who have tested at Ancestry and who wish to do more detailed comparisons is to upload their results to GedMatch, which is a free third-party website. GedMatch also have many additional useful features, such as a variety of admixture analyses, though there is a small fee to access some of the advanced tools. GedMatch also serves as a database where you can compare your results with people who have tested at all three testing companies, though the vast majority of people in the database are currently Americans. DNAGedcom provides a further range of tools for advanced users including Don Worth's wonderful Autosomal DNA Segment Analyser.

Now to get back to the subject of the AncestryDNA presentation at Back To Our Past. Mike Mulligan provided an introduction to autosomal DNA testing in a presentation that was pitched very much at a basic level. I was interested to learn that the Ancestry.com headquarters are in Ireland. In order to prepare for the launch, testing has been done on eighty people who are either Ancestry.com employees or their friends and family members. In contrast with my experience, Mike seems to have found quite a few close matches in the database. I suspect this is because a lot of Americans have quite recent Irish ancestry so people in Ireland are much more likely to get meaningful matches in the database. Mike has also had matches with a number of adoptees, and he's been corresponding with them and trying to help them. The price of the AncestryDNA test has not yet been decided. Mike mentioned that Ancestry did not want to give out segment data because of what they perceive as potential health issues.
Mike Mulligan of Ancestry.com
Mike showed us some examples of the ethnicity reports provided by AncestryDNA. I've included photographs of a couple of the slides below but unfortunately the lighting at the RDS was rather bright and, even with my best attempts at upping the contrast in Photoshop, I'm afraid the images are still rather over-exposed.
The most interesting slide was one which showed a comparison of the ethnicity estimates received by some of the Ancestry employees in Ireland who were included in the "friends and family" testing. Although it was only a very small sample it seemed apparent that the people who were from the north west of Ireland were the ones who came out with the highest percentages of  "Irish". Most of the Ancestry employees were between 70% and 100% "Irish" though one person was only 29% "Irish". However he had a Scottish great-great-grandparent and also some English ancestry.

I'm personally not convinced that the AncestryDNA test is reliably able to distinguish between "Irish" and "British" DNA. According to my Ancestry DNA test my percentages are:

47% Europe West
21% Great Britain
20% Ireland
8%   Iberian Peninsula
4%   Trace regions

I have one Irish ggg grandmother, one Scottish ggg grandfather, and all my remaining known ggg grandparents are from England. However, I do have some gaps in my family tree, including a London-born great-grandfather whose parentage is unknown and who is my biggest brick wall, so it's possible that I do have some more distant Irish ancestry of which I'm unaware, though not enough to account for such a high percentage. I also have a surprisingly low percentage of "British" DNA. I've found that lots of Americans come out with very much higher percentages of "British" than me!

Ancestry.com also had a stand at Back To Our Past and the displays provided information about the Ancestry DNA test. The stand was always busy when I went past but I was surprised to see them advertising the test without being able to offer visitors the chance to buy a kit.
The Ancestry.com stand at BTOP.
What's your story? Ancestry.com at BTOP.
On the Monday after Back To Our Past the ISOGG members who had helping out at the show were treated to a special day out which had been arranged with meticulous detail by Gerard Corcoran, the ISOGG regional representative for Ireland. (I wrote about our day out in my previous blog post.) Gerard had arranged for us to have dinner on Monday evening at Ka Shing, a Chinese restaurant in Dublin city centre. The evening was hosted by Ancestry.com, and three of the Ancestry team joined us for the dinner: John Halvey, Operations Manager, Ancestry International; Eric Booth, Senior Product Marketing Manager, International; and Mike Mulligan, the International Product Manager who had given the presentation on Saturday at BTOP. The evening provided us with an excellent opportunity to quiz the Ancestry staff about their autosomal DNA test. I was sitting next to Mike Mulligan. Mike had mentioned in his presentation that his DNA test had "confirmed" a genealogical connection with a sixth cousin once removed which had shown up as a shaky leaf hint. However, such a statement is highly misleading as it's quite possible that the shared DNA could have come from a different ancestor altogether. You can only verify such relationships by having access to the underlying segment data, and using techniques like triangulation and chromosome mapping to determine which ancestral couple contributed the matching DNA segment. For examples of how the AncestryDNA leaf hints have the potential to lead people up the garden path have a look at Heather Collins' blog post My AncestryDNA review: a cautionary tale and CeCe Moore's article AncestryDNA, raw data and Rootstech. Mike conceded that this is potentially a problem. He told me that Ancestry have done a lot of research and it is apparent that the vast majority of their customers are not interested in doing the advanced autosomal DNA analysis that is being conducted by some of the members of our genetic genealogy community, which is why Ancestry have not provided the tools. Their research certainly confirms my own findings from talking to people in the UK who have taken autosomal DNA tests. I find that very few people are up to the challenges of using the advanced techniques, and many struggle with the basics such as trying to download their list of matches into Excel. Intriguingly, however, Mike did mention that Ancestry have some sort of triangulation tool in the pipeline. He said he'd seen the results for one of his US colleagues using this tool and they were very interesting. We shall have to wait and see what materialises.
Dinner at Ka Shing courtesy of Ancestry.com. 
I didn't get round to discussing the discontinuation of the AncestryDNA Y-DNA and mitochondrial DNA tests. AncestryDNA stopped selling these tests in June this year, and their Y-DNA and mtDNA database was scheduled to be destroyed in September much to the horror of the genetic genealogy community. I understand that John Halvey, who was sitting on the other table from me, was getting a hard time from my fellow ISOGG members about the potential destruction of this irretrievable asset. However, CeCe Moore, who was at the bloggers' summit, has since advised me they were informed that the "legacy" database has not been destroyed after all. She tells me  "It's fate remains up in the air for now, but fortunately all hope is not lost."

Competition is very healthy, and it will be interesting to see how the genetic genealogy market evolves in 2015.

This post was updated on 13 November to correct the information on the AncestryDNA Y-DNA and mtDNA databases following a comment received from CeCe Moore.

See also my blog post dated 8th January 2015 - The Ancestry Y-DNA and mtDNA samples have not been destroyed after all.

© Debbie Kennett 2015

7 comments:

-rt_/) said...

Completely irrelevant but a Chinese restaurant in Dublin makes me think "How appropriate the cultures meet again".
The first American transcontinental railroad was built by Irish and Chinese workers. The Irish started in the east and worked westward; the Chinese in the west going east. They met in Wyoming.

Your Genetic Genealogist said...

Hi Debbie,
Great summary. I just wanted to mention that at the blogger summit we attended at AncestryDNA last month, we were informed that the "legacy" database has not been destroyed. It's fate remains up in the air for now, but fortunately all hope is not lost.

Debbie Kennett said...

CeCe, Many thanks for setting the record straight. That is very good news. I've updated my blog post to reflect the correct situation.

Kennehy said...

Hi Debbie,
Thanks for the summary – love your blogs. Once again this shows that genealogists in Britain (and the rest of Europe) should be targeted by companies such as FTDNA and Ancestry to address the imbalance of a US-centric DNA database. Yes I know they will only do it if they can determine it is commercially viable, but advertising the benefits and using all opportunities (like family history shows) to sell tests will confirm it is doable.
I live in Australia but my ancestors are British and I find it frustrating tracing links from 3-5th cousins who migrated to the US many centuries ago and whose tree is limited or non-existent. It just doesn't work for me. At 67 and 111 markers I still cannot get a close (and usable) YDNA match.
I encourage as many Europeans and their immediate descendants to take DNA tests - the results of which can, at least, be stored for future generations who may be more interested in this type of genealogy tool.

Ken

Debbie Kennett said...

Hi Ken, Thanks for your kind words. I agree it's frustrating having all these matches with distant US cousins with limited or non-existent trees. Even when they do have detailed trees it's impossible to find the connection.

Family Tree DNA have been actively promoting their tests in the British Isles in the last few years by attending Who Do You Think You Are? Live and Back To Our Past. There are lots of people in the UK who are actively involved in DNA Projects (eg, many of the UK members of the Guild of One-Name Studies). iGENEA, FTDNA's European partner, also bring in many European testers.

One of the problems is that the Americans with Colonial Ancestry are effectively an endogamous population all tracing back to a few thousand people. They all share multiple ancestors in common. This all means that they get many more matches. Those of us with British ancestry go back to a population of several million just a few hundred years ago so we need proportionately many more people in the database to get the same results as our American friends.

Mountain Mama said...

Thanks for your concise summary. My big concern is that Ancestry DNA, as it stands, is totally worthless for breaking down brick walls unless the potential matches will upload to FTDNA or GedMatch. Do I match the potential wall-breaker on a significant string that triangulates with others of that surname? Is the match with some other distantly related ancestor? Is the "match" too small to be significant? No way to tell so one is left jumping to conclusions and you know how accurate that is.

Debbie Kennett said...

Hi MountainMama

I agree that you can't draw any conclusions from AncestryDNA matches with the system as it currently stands. It's a particular problem for people with Colonial American ancestry and Ashekenazi Jewish ancestry who are invariably related to each other on multiple ancestral pathways. The AncestryDNA test can currently really only be used for very close relationships up to about the third cousin level where both parties have fully documented pedigrees and know for sure that they are only genealogically related on one line. Otherwise, as you say, you need to get your matches to upload to GedMatch or Family Tree DNA if you wish to verify the connection. We'll have to wait and see how the new triangulation tool works out when it's launched.