View Full Version : Ancestry Composition compared to actual matches

12-09-2012, 03:36 PM
Well the speculative version of Ancestry Composition seems the most accurate for my mother. She is half UK half unknown American and gets:
59.5% British/Irish
11.6% French/German
0.8% Scandinavian
26.3% non specified North European
0.9% non specified South European
0.7% non specified European
0.1% East Asian/Native American
0.1% Sub Saharan African

I was disappointed to see that Eastern European did not show up. I say this because my mother gets 10 cousin matches to Hungary across 23andme and Gedmatch. 3 of these share a common ancestor with her and they have been able to locate surnames and a remote town in Transylvania in common. They are Szekely Hungarians. Mum also matches one of these cousins on two different chromosomes and shares these segments with the cousins son and Granddaughter. They all appear strongly Eastern Euro on AC.
10 matches to Hungary seems more than coincidence so why hasn't it shown up? The other 5 matches are all over the threshold for IBD matches. All the Gedmatch calculators hint at minority Eastern European. She gets other Eastern Euro cousins too.
I have seen results of someone who is 6.25% Croatian with ancestry from a port used by Hungarians, the rest British and she only gets 0.4% East Euro and 0.7% Balkan but gets 29% non specified north Euro, so could her Croatian and my mothers Hungarian be hidden in this category?

12-09-2012, 03:45 PM
Yes, it could certainly be hidden there. There are probably not as many samples from these places in the database, so there may be a tendency to overreport data from very well represented populations, and underreport data from poorly represented ones.

12-09-2012, 03:58 PM
Thanks. I did see they use about 68 people from Hungary as part of the East Euro group, but maybe that is not enough samples to represent a whole country? I also see they have no Central Euro group but Switzerland comes under Northern Europe so maybe that would affect things too?

12-10-2013, 09:18 AM
One thing that should be noted is that the changes have not occurred all at once. As I've logged in recently, I've been getting different results every few days.

As of today, my "Southern European" is at its lowest point so far. It's down to just 1.4% in the standard estimate, of which 0.3% is Iberian and the rest is nonspecific Southern European. This actually represents a tiny increase to Iberian, from 0.1 to 0.3%, but this is more than offset by decrease in the nonspecific part -- from 1.9 to 1.1%.

But it's in the speculative estimate that the biggest change can be seen. Previously, I had 2.2% Italian, 1.3% Balkan, 1.0% Iberian, and 4.1% nonspecific Southern European. This made a total of 8.6%. Now, I have slightly more Iberian (1.7%), but less Italian (0.8%), and no Balkan. Nonspecific Southern European has dropped to 2.4%, so my new total is only 4.9%.

This is in the face of my maternal grandmother having been half Spanish, with both of her grandfathers actually having immigrated from Spain. (Separately, from the island of Menorca. There is no evidence they knew each other.)

By contrast, my father -- who has no known Southern European ancestry -- formerly had 2.2% nonspecific Southern European in the standard estimate, and 3.8% in the speculative estimate (nonspecific). Now, he has 1.3% Italian, 0.3% Balkan, and 3.9% nonspecific Southern European. This means that at 5.5%, Ancestry Composition believes my father has more Southern European ancestry than I do.

Further, while "split view" is again working, it is again also giving peculiar results. For example, in the standard estimate, I previously had 1.8% East Asian and Native American. 1.4% of this was Native American, and the remainder was nonspecific EA and NA. My father had 0% for EA and NA, and "split view" reasonably reported that ALL of my EA/NA ancestry was maternal. In the speculative estimate, I had 2.0% East Asian and Native American, which was essentially all reported as Native American. (Presumably due to rounding, 2.0% was reported as NA even though 0.1% was also reported as nonspecific.) Again, this was shown as all maternal in split view.

Now, my father still shows no East Asian and Native American in any estimate. However, in split view 0.2% of my EA/NA is nevertheless shown as "paternal" in the standard estimate, and 0.3% in the speculative estimate.

So I inherited EA/NA from my father that he himself does not have, though no southern European. Incidentally, one of my sisters did manage to inherit around the expected 12% southern European ancestry (based on paper trail). However, she only inherited some of that from our mother. 0.4% Balkan and 1.8% nonspecific is from our father.

Curiously, in chromosome view my father's 0.3% Balkan all appears on chromosome 8. But in my sister, it all appears on the X chromosome. So this Balkan DNA has managed to both migrate from one chromosome to another, and to increase by 1/3. Clever little SNPs, I'd say. :-)

EDIT: Here's yet another curious little bit. In the speculative estimate, my wife is 65.3% British and Irish, 12.3% French and German, and 17.3% nonspecific northern European. This is all believable so far. Then she has 1.6% Italian, <0.1% Sardinian, and 2.3% nonspecific southern European. She has no known southern European ancestry, but it's still in the realm of possibility. And this is all capped off with 1.2% nonspecific European.

The curious part is just this. With both parents genotyped, phasing our daughter's chromosomes should be relatively easy. Yet she reportedly has 0.3% Middle Eastern that neither parent does. (0.1% North African, and 0.2% nonspecific ME.) In split view, it appears this ME is from me -- even though I have no such ancestry myself, says Ancestry Composition.

In my daughter's chromosome view, all of this ME is shown as an unbroken segment on chromosome 4 (though part of the segment is North African). For my own copies of chromosome 4, one is identified as fully British and Irish. The other, however, is partly southern European (including Italian, at the beginning portion). This appears to extend into, but not through, the region that in my daughter is identified as ME. The remainder of the corresponding region is (in me) tagged as "nonspecific European".