Playing with percentages

The numbers game

So despite the fact that The Legal Genealogist keeps saying, over and over, that the admixture percentages we see in all the autosomal DNA test results are only good for cocktail party conversation, the fact is, hey, at this time of year there are a lot of cocktail parties.

And, of course, a lot of admixture tools. One of ‘em just got updated, and I couldn’t resist taking a look.

K13The tool is the Eurogenes K13 admixture tool; it’s accessed through my favorite third-party utility site, Gedmatch.

You head over to the site, log in, choose Admixture from the Analyze Your Data options, select Eurogenes from the drop-down menu and “Admixture Proportions (with link to Oracle)” from the radio button choices, and then click on continue.

Enter your kit number at the top, select Eurogenes K13 from the drop down list for calculator models and click continue again.

And then you wait while it runs through 1000 iterations comparing your raw data to the data in its population samples using its algorithms.

The first thing I wanted to see was how my mother’s two brothers and one sister whose kits we have at Gedmatch compare to each other. It was intriguing to see how their results are close — but not exactly the same.

Eurogenes K13 Admixture / Percentage
  Uncle 1 Uncle 2 Aunt
1 North Atlantic 48.58 North Atlantic 46.72 North Atlantic 48.03
2 Baltic 23.33 Baltic 23.11 Baltic 25.31
3 West Med 13.43 West Med 13.22 West Med 12.63
4 West Asian 6.75 East Med 6.83 East Med 5.1
5 East Med 4.27 West Asian 5.7 West Asian 4.94
6 Red Sea 1.33 South Asian 2.18 Red Sea 1.23
7 Northeast African 0.98 Northeast African 0.99 South Asian 1.06
8 East Asian 0.47 Red Sea 0.74 Oceanian 0.52
9 South Asian 0.44 Sub-Saharan 0.32 East Asian 0.5
10 Oceanian 0.34 Oceanian 0.18 Northeast African 0.41

You can see from that table that the top five general areas that make up the admixture of all three are the same — but not in the same order and not in exactly the same percentages. (And, by the way, everything under one percent is pretty much just noise.)

Then we can match them up against the specific sample populations used by this tool to see which modern populations they’re most like. And, again, they’re close but not exactly the same:

Eurogenes K13 Single Population / Distance
  Uncle 1 Uncle 2 Aunt
1 Southeast English 3.28 German 3.54 Dutch 2.34
2 Dutch 3.46 Southeast English 4.23 German 3.29
3 Orcadian 3.99 Dutch 4.86 Southeast English 3.41
4 German 4.16 Danish 5.66 Danish 3.7
5 Southwest English 4.19 Orcadian 6.01 Orcadian 4.09
6 Danish 4.22 Southwest English 6.3 Irish 5.26
7 Irish 4.48 Irish 6.99 Southwest English 5.41
8 West Scottish 5.04 West Scottish 7.36 West Scottish 5.77
9 Norwegian 7.53 Norwegian 8.82 Norwegian 6.19
10 Swedish 8.52 French 9.23 Swedish 6.94

Then I wanted to look at a more personal comparison. The comparison between me, 50% German and 50% colonial American mutt, and my half-brother, 50% German and 50% Swedish (all four of our father’s grandparents were born in Germany and all four of his mother’s grandparents were born in Sweden).

I wanted to see how well this particular tool discerns the German and the Scandinavian in that sort of mix since all of my German results disappeared in the most recent AncestryDNA admixture analysis, And here’s what we got:

Eurogenes K13 Admixture / Percentage
  Half-sister (me) Half-brother
1 North Atlantic 44.55 North Atlantic 47.3
2 Baltic 26.14 Baltic 30.8
3 West Med 12.93 West Med 8.67
4 East Med 6.22 West Asian 6.29
5 West Asian 5.48 East Med 3.81
6 Red Sea 1.84 South Asian 1.27
7 South Asian 1.62 Red Sea 0.89
8 Northeast African 0.95 Amerindian 0.84
9 Amerindian 0.27 Oceanian 0.07
10 Sub-Saharan 0.07

Those are the overall numbers — now for the populations:

Eurogenes K13 Single Population / Distance
  Half-sister (me) Half-brother
1 German 2.23 Swedish 3.02
2 Dutch 4.46 Norwegian 4.14
3 Southeast English 6.35 Dutch 5.51
4 Danish 6.8 German 6.26
5 Orcadian 7.23 Danish 6.88
6 Norwegian 7.91 North Swedish 7.44
7 Swedish 8.04 Orcadian 7.83
8 Irish 8.26 Irish 8.37
9 Southwest English 8.35 Southeast English 8.95
10 West Scottish 8.94 West Scottish 9.18

So… this admixture tool spots our German (which appears to look a lot like Dutch), and spots my brother’s Swedish (which appears to look a lot like Norwegian).

There are two things I find interesting about this tool, since we’re chatting now over our cocktails.

The first thing is that, unlike the recent change in the analysis by AncestryDNA (see DNA disappointment, posted 15 September 2013), this admixture calculator doesn’t seem to have any trouble finding the Germans. I get ‘em, my brother gets ‘em, even my mother’s siblings get ‘em (and that makes sense — we have a Shew line that I expect goes back to some German Schuhs).

And the other thing is how very alike these reference populations must be for the results to be coming out the way they are. German and Dutch and Danish all so very close, Southeastern and Southwestern English and Irish and Orcadian.

I still want a magic wand I can wave that would put an ethnic and geographic origin stamp on every segment of my DNA. For that matter, I want a magic wand that will mark each segment with the name of the specific ancestor I inherited it from — maiden names of women included.

But in the interim, it’s really fascinating — here around the Sunday cocktail table — to see that we’re really not French or German or Dutch or Danish.

We’re simply human.

And so much more alike than we are different.

Print Friendly
This entry was posted in DNA. Bookmark the permalink.

7 Responses to Playing with percentages

  1. John Tracy Cunningham says:

    Judy

    Agreed that admixture analysis is far from settled! My results:
    FTDNA FF – 94% from the Pyrenees region, both sides, plus the Orkneys (which apparently were settled by Iberians after the ice melted last time); 6% Tuscan/Sardinian
    23andMe – 100% European, not much definition after that; not very useful
    Eurogenes K13:
    - 48% North Atlantic
    - 23% Baltic
    - 13% West Med
    - 8% East Med
    - 5% West Asian
    Oracle Single Population Sharing:
    - Southeast English, German; then
    - Dutch, Danish, Orcadian, Southwest English; then
    - Irish, West Scottish; then deeper down
    - Dash of Hungarian
    - followed by a whole collection of Iberian and Southwest French (Pyrenees again)

    My known ancestry is English, German, Irish, and Scottish, and there may be a Frenchman back there somewhere. DNA points to the Pyrenees (mtDNA V). I wonder if FTDNA is trying to tell me that all of my ancestors came out of the Iberian refuge and spread up the west coast of Europe?

    Regards

    John
    Stuttgart

  2. Fascinating numbers. I am Dutch going back 10 generations, and so is my mom. We both tested and in both our cases, German is the nr. 1 reference population, with Dutch a close second. I think there’s not much difference between Dutch and German, genetically speaking.

    What I find hilarious is that my father’s phased kit (he hasn’t tested) comes up 3% Amerindian (I’m at 1.5%). I’ve been able to trace 10 generations of his ancestors on all lines except 2 or 3 people, and they all come from within 10 miles of the house where he was born, in a village on the Dutch/German border. Definitely no Indians there! I’m sharing this because a lot of Americans get excited when they see 3% Native American and think this cannot be a statistical anomaly. But believe me, if my father can show up 3% native, it can definitely be a fluke.

    • Judy G. Russell says:

      I suspect that Dutch and German (and probably Danish as well) will prove to be just about indistinguishable in the long run, Yvette. Now as for that Native American — some interesting recent developments suggest that the even more ancient ancestors of our Native Americans may have “fed” their DNA to Europe as well as the New World. We have more to learn there too!

  3. Paula Williams says:

    Um, Baltic?

    Mine are fairly close to the aunts & uncles. Interestingly, my West_Med is identical to our Aunt’s.

  4. Shirley Ann Rankin says:

    I’m confused by the Single Population / Distance figures. What does “distance” refer to?

    • Judy G. Russell says:

      It’s a measure of the differentiation between you and the reference populations. The higher the number, the more different you are.

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>