DNA disappointment

Who do they think I am?

Once upon a time, when The Legal Genealogist was a very small girl,1 the Hot New Toy for the Christmas season was a particularly frilly baby doll. Blue eyes, short blond curls, frilly baby clothes, the works.

girlThat was the year when I didn’t care much what all I got for Christmas … as long as one package, any one package, contained that doll.

And that was the year when my mother didn’t get the Sears order in anywhere near early enough and, by the time she ordered, the doll that was left — the doll that ended up in my stack of gifts — was a Raggedy Ann.

Now don’t get me wrong. In a family the size of mine, any gift is a good gift. And a Raggedy Ann is a very fine doll, one that at another time perhaps would have been loved and treasured.

But to my five- or six-year-old self, it was a bitter disappointment.

That pretty much sums up my personal reaction to the new AncestryDNA Ethnicity Estimates that were rolled out to the first group of customers on Thursday.

Because, you see, many of the people included in that first group — all customers will see the new estimates within the next month or so — they got the baby doll.2

Me… I got Raggedy Ann.

Unlike many AncestryDNA customers who thought their old ethnicity estimates were way off, I thought my old ones were probably pretty close, except for what I thought was a high estimate of Scandinavian since I have no known Scandinavian ancestry. They seemed to reflect pretty well my 50% German ancestry from my father and my mother’s more mixed but largely British and Scots-Irish (not Irish, mind you, but Scots-Irish, which isn’t Irish at all) heritage.

The new estimates… well…

ADNA.comp

So… okay… where did my Germans go? And where did all those blankety-blank-blank Scandinavians come from? And why, oh why, is everybody else getting ethnicity estimates that more closely approximate their known ancestry — and I’m getting estimates that are farther away from mine?

Why does National Geographic’s Geno 2.0 say the population I most closely resemble in today’s world is the German population — and AncestryDNA shows me with just a trace amount of DNA from what it now calls Europe West (Belgium, France, Germany, Netherlands, Switzerland, Luxembourg, Liechtenstein)?

Why does 23andMe report much higher percentages of French and German, and much lower percentages of Scandinavian, no matter which level of analysis I look at there?

And beyond the German-or-Scandinavian results issue, why does 23andMe say it can detect trace amounts of Sub-Saharan African in my ancestry — and AncestryDNA says the African it’s picking up is from the Saharan region (called Africa North: Morocco, Western Sahara, Algeria, Libya)?

Now I’ve carefully read the Ethnicity Estimate White Paper dated 4 September 2013 that AncestryDNA offers to its customers who see the new results.3 It explains, carefully and in detail, how each of our DNA samples is compared to those of what are called reference populations: groups of people from specific geographic areas whose pedigree establishes that their great grandparents, grandparents and parents were also from those specific geographic areas.

And there are parts of this White Paper we all need to remember:

• “When considering AncestryDNA estimates of genetic ethnicity it is important to remember that our estimates are, in fact, estimates.”

• “(I)n genetic ethnicity estimation, we are attempting to estimate the unknown amount of DNA actually inherited from all of a sample’s ancestors. Genetic estimates of ethnicity also go back thousands of years, beyond the end of a pedigree paper trail. Regions identified as ‘populations’ in a pedigree may have been very different thousands of years ago, and so may be represented differently in a genetic ethnicity estimate.”

• “Britain and Western Europe are geographically close, and a significant amount of historical migration (and hence, interbreeding) has occurred between these regions. Frequent interbreeding has led to very little genetic differentiation between these two regions, and thus our current approach has less power to identify the true ancestral source for individuals with ancestors from these locations.”

So why — given the fact that I know darned good and well that these are (a) estimates, (b) estimates based on comparisons not to actual historical populations but rather to small groups of people living today, (c) estimates based on comparisons not to actual historical populations but rather to small groups of people living today and based purely on the statistical odds that those small groups tell us something meaningful about past populations — why am I disappointed?

This may be a perfectly good set of results. Maybe all of my Bremen ancestors (documented back many generations thanks to civil registration and church records) actually descend from a pocket of displaced Hanseatic League sailors from Stockholm or Bergen.4 Maybe a bunch of the British and Scots ancestors in my tree descend from invading Vikings.

The problem is… we don’t know what these statistical analyses of our genes today are really telling us about our ancestors who lived so many years ago — decades, generations, perhaps millennia in the past. Maybe we’ll never know. Particularly for those of us with nearly 100% European ancestry, comparing ourselves as living beings today against other living beings today may never tell us with certainty what we’d like it to tell us.

What I want, of course, is a road map to every part of my genome: this part is English, this part Scots, that batch over there German and that part French. What I want is for all my DNA results from all the companies to agree with generally conforming percentages rather than being, literally, all over the map. What I want is for the science to be there, right now, today (yesterday would have been better) to take all the uncertainty out of this.

What I want is the baby doll.

And, once again, what I got was Raggedy Ann.

Darn it all.


SOURCES

  1. And when dinosaurs roamed the earth…
  2. See, for example, Blaine Bettinger, “AncestryDNA Launches New Ethnicity Estimate,” The Genetic Genealogist, posted 12 Sep 2013 (http://www.thegeneticgenealogist.com : accessed 14 Sep 2013).
  3. AncestryDNA, Ethnicity Estimate White Paper, 4 Sep 2013, online to select AncestryDNA customers (http://dna.ancestry.com/ : accessed 14 Sep 2013).
  4. Wikipedia (http://www.wikipedia.com), “Hanseatic League,” rev. 29 Aug 2013.
Print Friendly
This entry was posted in DNA. Bookmark the permalink.

36 Responses to DNA disappointment

  1. I’m still waiting to receive my revised ethnicity results, and I shall be interested to see what I find. However, I was amused to discover that, according to Ancestry, Britain is no longer part of Europe!

    • Judy G. Russell says:

      I suspect there’s a fair number of folks who wish it wasn’t, Debbie! But keep in mind: this is playing to a largely US audience meaning lots of folks who want to know English versus Irish versus all those continental types.

      • I now have my updated ethnicity results. Like you I found my results very disappointing. I live in England, and all my ancestry as far back as I can trace it is from the British Isles. Yet according to Ancestry’s new ethnicity estimates only 21% of my DNA is from “Great Britain”. According to Ancestry you’re much more British than me! Ancestry have also given me far too much “Irish”. I’ve written a blog post where I share my thoughts as to why I think Ancestry have got it so wrong:

        http://cruwys.blogspot.co.uk/2013/09/my-updated-ethnicity-results-from.html

    • betty says:

      um…those results don’t indicate that GB is not a part of Europe. It shows Europe and then two subcategories: GB and Scandanavia. What are you even talking about?

      • Judy G. Russell says:

        Um… you might want to read the piece again, and then maybe again. The issue is the inability of AncestryDNA to detect what it’s now calling Europe West (Belgium, France, Germany, Netherlands, Switzerland, Luxembourg, Liechtenstein). Not Europe overall, but Europe West. If you look at the chart on the left, you see that AncestryDNA previously detected 43% of my genome from Central Europe — presumably my paternal German ancestors — and 50% from Great Britain and Scandinavia — presumably my maternal Scots-Irish ancestors. The newer results, on the right, have the Central European segment disappearing almost completely.

  2. Mary E Hall says:

    Great summary, Judy. Throw in the various sadmixture results at Gedmatch and the whole DNA ethnicity “industry” seems almost like palm reading. I know it’ll get better.

    Looking forward to seeing my (and my mother’s ) AncestryDNA results. Right now, she has way more Scandinavian than I do, despite the fact that my paternal grandparents were FROM Denmark….and she has NO known Scandinavian roots in her 10 generation pedigree filled with early New Englanders and Palatines.

    Still think 23andMe makes the most sense, right now.

    • Judy G. Russell says:

      We can certainly HOPE it gets better, Mary, but remember the limitations: we’re comparing to living people not historical populations.

  3. Judy,
    My Germans went the same place as yours—partly covered in Great Britain and partly covered in Scandinavia where those larger concentric circles overlap over Germany. My hubby on the other hand got his to behave and they are in the category West Europe….where I think we all would have been happier if they landed there. But as bugged as I was in the beginning I do think the way it is presented, the level of detail and the ranges given make for a vast improvement over last time. I no longer have 9% uncertain.. My Caucasus Mtns, African etc are where they should be. My colonial British is now British. As I explained on the thread at 23andMe the problem isn’t with the science as much as it is with the advertising claims of the companies and our expectations based on thiee claims. As I mentioned there the companies are wising up now using “speculative mode” and “ethnicity estimate.”

    This science needs much larger samples and much more work but as long as one views it as in its infancy then we can be realtively pleased with the changes. The idea that they can come even close is pretty cool when you start looking at individual AIMs

    • Judy G. Russell says:

      Sigh… I’m glad to hear I’m not the only one with disappearing Germans!

      As I said in Friday’s blog post (AncestryDNA begins rollout of update), there’s quite a bit to like about the new estimates and the amount of data provided is great. But all the samples in the world may never be enough to overcome the simple fact that we are looking at modern populations and trying to extrapolate backwards based on statistical analysis of what we think we know about the last three or four generations.

      Moreover, the public focus on these percentages as if these were the be-all-and-end-all of autosomal testing is so misplaced for genealogy. In most cases, these percentages don’t mean a thing in terms of what we really want from autosomal testing: cousins with the answers to questions we can’t answer!

      • Judy,

        “Misplaced” is right! I try to go into this in lesson 7 where I say “Each of these great grandchildren has the same great grandparents but they did not inherit the same DNA. ”

        What is interesting is that Ancestry is using 300,000 AIMs up from 30,000 and the sample size has gone from about 3,000 to 4,245 and they have increased to 26 global regions. I woukd say that when they get to 1,000 per each global region then we will see a tightening up of the AIMs. Interesting that the Irish DNA Atlas Project needed eight great grandparents all born within something like 15 kilometers of each other. If that sort of thing was replicated across the globe then we would be getting somewhere.

        However the human diaspora is something we just don’t accept as if it is counterintuitive. Germans are probably among the worst since they were sometimes Germans and sometimes French and sometimes Dutch, sometimes Polish, sometimes Swiss and sometimes Austrian as in my maiden name line who were probably from the Caucasus and then Austria and then to Bavaria in the 1600′s and then to PA in the early 1700′s and then to VA, NC, TN then AL. And then not liking to stay put, appraently being in their blood they moved back up to TN then to MO and IA. And no they weren’t done yet. Then to Nebraska and out to Colorado and then to California! So if I had to assign a nationality what should I pick? And that’s just one line! Therein lies the conundrum.

        • Judy G. Russell says:

          >> what should I pick?

          Mutt for me! Or Heinz 57 varieties!

        • Nathan Machula says:

          Kelly, the 300,000 SNPs mentioned toward the beginning of the white paper was just a starting point—the intersection of SNPs between the Illumina OmniExpress used by Ancestry and the Illumina 650K that genotyped the HGDP reference samples. Later in the white paper, Ancestry said they “look at over 100,000 highly informative SNPs.”

  4. Annick says:

    Judy,
    My DNA has not been tested in anyway, so I don’t have any horse in this race! But I am from France and recently studied my region of Lorraine during the 7 years war. The German Emperor used loads of Scandinavian troops to fight in that war and they ravaged that part of the country, as well as Alsace and many Lander in what is now Germany. Do you think there could be a correlation between that episode in the 17th Century and your DNA showing such a high percentage of Scandinavian ancestry? Just a thought!

    • Judy G. Russell says:

      That’s certainly one possibility, Annick, along with the Hanseatic League sailors (both Bremen, my father’s birth city, and many Scandinavian cities including Stockholm and Bergen, were Hanseatic League cities so there was a lot of trade back and forth). But it points up the general difficulties of trying to extrapolate from modern populations back into historical time.

      • Delores says:

        I am studying to become a anthropologist; therefore I have taken several classes on the study of humanity. I agree with the problem with modern dna test results. People have always moved around the several continents since pre historic times and even before then. If you go back far enough the earth use to be one land mass and the people at that time have to have their dna in all of us. dna mutate over long periods of time, but we have people that are testing positive for Neanderthal dna(not included in 23 and me or Ancestry.com) Also, the way dna works, is that it tumble like the lottery, meaning that you et 50% from your mom and 50% from your dad, but each child does not get the same combination. Therefore we can get a lot of different results in members from the same family. We all have two ancestries. Genotype (hidden) and Phenotype (traits that are seen, like blue eyes.

        • Briana Felch says:

          Just to clarify, Neanderthal DNA IS indicated in 23andme, just not in the admixture screen where the rest is indicated. It is under a different screen in Ancestry Tools. (And it has been there since my results came in in Feb. 2013).

          I take it to be the separation from the regular admixture is that perhaps that is more speculative data or because that 2.7% (in my case) cannot perhaps be easily sorted out from say my British or German or Iberian or whatever (and may overlap several admixtures). In other words, per my admixture which roughly equals 100%, adding the 2.7% would also be confusing to some as it would appear I would have 102.7% DNA. I assume also as well that some people might not take too kindly to hearing they have Neanderthal DNA, so this could be another reason it is separated out. But that is just an uneducated case on the why of that. I don’t have FTDNA as a comparison for how they do it.

          I have only tested at Ancestry (you are right, no Neanderthal or Denisovan there) and 23andme, but just ordered my dad’s y-DNA test from FTDNA to learn about my paternal Smiths. I have done no testing thus far at FTDNA or Geno. Not sure I will (besides Dad’s and perhaps husband’s y-DNA at FTDNA). If money allows, maybe in the future.

          Here is my info from 23&me on Neanderthal. Tried to copy the 23&me logo as further evidence, but it wouldn’t copy.

          ANCESTRY TOOLS NEANDERTHAL ANCESTRY
          DISCUSS
          SHARE
          RATE
          This lab estimates your genome-wide percentage of Neanderthal ancestry

          Got Neanderthal DNA?

          An estimated 2.7% of your DNA is from Neanderthals.

          Briana Felch (you)
          2.7%
          44th percentile

          Average European user
          2.7%

          Just thought I would share that as some readers may not be aware of this and may wish to know.

  5. Is there really a difference in German, French or English ancestry? The royal family in England had cousins that were rulers in Germany and Russia. They looked very similar to each other. Don’t know if they were in the French royal family. My grandmother on my father’s side was part-German, but she was also part-Scotch-English. Some German words were still being used when I was growing up in the 1950s.

    • Judy G. Russell says:

      There’s undoubtedly some difference, David — especially when you’re NOT talking about closely-inbred royal families. The real question is, what is the difference and can it be detected?

  6. Kat says:

    I can sympathize about Raggedy Ann. When I was about 6, I wanted a Margaret O’Brien doll and Sears sent a generic baby doll!
    I’ve not submitted to DNA testing, but believe there is room for improvement in the interpretation of results. For the present, I have more known relatives than I can keep up with.

  7. George Clark says:

    Sorry your results were not better. Both my new results and my father’s new results were close to our ancestry from our paper trail.

  8. Paul Caverly says:

    Although I am no expert I have used genealogy DNA for 10 years in my small DNA project. It did help dispel one of our surname folklore that existed for over 100 years and created 3 distinct genetic trees for are not so common surname. We utilized FTDNA for this study but recently I also did a test at 23andme. Previously I also had the ‘FamilyFinder’ test at FTDNA and it was interesting but finding common ancestors is a challenge. Using results from both sites I did find a 4th cousin.
    Now I am trying to search a more difficult family tree surname back to N. Ireland. I am utilizing a case study from http://www.irishorigenes.com for the surname HENRY. The study provided more clues for searching Irish records in a geographic area for the pre-1850s. The case study did support some of my earlier guess as to where my ancestors may have lived. Now, to see if we can find any documentation to support the case study.
    I hope to attended your OGS webinar in Oct.
    Paul R. Caverly
    Canada

  9. Jade says:

    Judy I love your columns and your work in general, but I hope you will forgive a cautionary note.

    We have always to be aware of the impact of historical migration / invasion / trade patterns and how they can impact ancestries. For example, among children of ostensibly Chinese ancestry there are occasional births in Hong Kong of red-haired children reflecting 16th-century Dutch heritage.

    If one does a bit of reading on history of the so-called Vikings, one sees that the “Rus” from present Finland had a couple of generations of strong influence and a small empire centered on present Ukraine before the 11th century CE, and their name is now reflected in “Russia.” Their taking slaves and selling them from the ‘Slavic’ population gives us our term ‘slave.’ The ‘Vikings’ from present Denmark, Sweden and Norway also had their own impacts, such as in Ireland, and as far as the Mediterranean. Then subsequently with the “Mongol” invasion of Europe as far as Finland, Hungary and Austria, came bloodlines from central and east Asia. So there are a number of ways one could get “Scandinavian” and “Asian” ancestries in comparison with populations with like heritage, regardless where their ancestors had lived in the most recent few generations for which paper trails are a lot easier than back to medieval and earlier times.

    As you yourself have noted, the genes kept by each individual from mtDNA will change as passed down over time, and it can be a fluke that one lineage keeps elements from Viking and/or Slavic heritage that may not be reflected easily in paper trails.

    • Judy G. Russell says:

      Jade, you are absolutely right — which is why I’m NOT saying that these results are wrong. What’s clear is that we don’t know enough yet to know whether they’re right or wrong, and we may never be able to know enough.

  10. Bobbi says:

    Don’t forget that the Swedes entered the 30 Years War and were a presence in Northern Germany.

  11. Karel says:

    The geographical composition of our genes will tell a very different story depending on how far back we go. This means someone will be 100% British (or US or whatever) in the last generation. 10 generations ago, you will be something else than 100 generations ago. Etc. etc. And it is still the same person. There were times like last glacial maximum where all people were 0% British or German or Czech.
    My own results say basically the same. My ancestors are 100% from South Moravia (Czech Republic) in the last 150 years, all speaking Czech, but they are from South Moravia and Eastern Bohemia in the last 300 years, speaking Czech and German. According to Genographic, I have 2% of Native American DNA next to 43% North Europe, 36 % Mediterranean, 18% SW Asian. (No wonder this is quite similar to German or British reference populations). My Y-chromosome is N1c1 and mtDNA is H3. While the outer lineages (mother-mother… and father-father…) tell me that I am basically something like Finnish-Basque, my parents have no clue their lineages lead to these nations and only speak Czech. Between those two lineages, I can be 2% Native American as some ancestors certainly came from Siberia 10,000 years ago or more (N1c1 confirms that). It all depends how far we go and we get a different results and it would still be correct I think.

    • Judy G. Russell says:

      You are absolutely right about that, Karel, and I think that underscores why we can’t take these numbers all that seriously. At least not yet!

      • Brenda says:

        Not to mention that before 1620,,English and Spaniard’s were coming over here stealing natives from the new world and taking them back to Europe as slaves.I also found out about some infidelities within my immediate family that I never believed it would have happened since my folks were supposedly very good Christian’s.My paper trail goes to William the conquer.All my life I thought I was mostly British till the line went further into France.Just about all my English lines changed in the 1400′s to French,I recently took the ancestry dna but dont have the results..hey why dont they exhume some of the bones from then to get actual readings? They did it on one of our presidents.

  12. Thilo Agthe says:

    The movements of the Swedish army during the 30 years war were not confined to northern Gernany. The Swedes ranged through Saxony and also touched Thuringia. Several major battles took place in the area.

    So don’t discount the possibility of Swedish ancestry.

    Regards,

    Thilo C. Agthe

    • Judy G. Russell says:

      I’m not ruling it out — but the YDNA haplogroup of E-V13 makes it less likely that the Thuringen line was Swedish. Not impossible, just less likely.

  13. Pingback: Ethnicity Results – True or Not? | DNAeXplained – Genetic Genealogy

  14. Pingback: 2013’s Dynamic Dozen – Top Genetic Genealogy Happenings | DNAeXplained – Genetic Genealogy

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>