With 23andMe’s new update to its ancestry, the results for my family have changed. Not for me, since I’m not of European descent, and this looks Euro-focused (no surprise). But my wife and kids are different.
My wife has two great-grandparents who were born in Norway. 23andMe is picking that up immediately. It also picks it up in my children, from left to right, my daughter, my younger son, and my older son. With more than 3 million in their database 23andMe has knowledge of which haplotypes are unique to Norway, and which are not. When you click “Norway,” it says “We predict you had ancestors that lived in Norway within the last 200 years.” That’s telling me that they detect IBD segments uniquely found in Norwegian populations of a particular length threshold.
My youngest is on a new chip, so the Western Asian & North African I dismiss. But I’m not sure I believe some of the European admixture estimates. The two boys exhibit very little drop off in Scandinavian. But my daughter is way lower. This is not unreasonable, but they also exhibit differences in East Asian ancestry. And I’ve looked but I can’t detect this on PCA plots. My daughter is, in fact, more distant from Han Chinese than my sons.
In the future, I think perhaps genealogy-focused results, which show matches within particular nations, should be partitioned from admixture analysis. That’s how it used to be.
(it is a curious coincidence that both my more Scandinavian children are heterozygotes on the KITLG locus for the derived variant, though I know they get it from their mostly German grandfather)
“200 years” is boilerplate. I have ancestry from two countries in one region, and on the “details” spreadsheet, it shows highest-confidence to both of these countries as well as two more nearby nations (5 on a zero to 5 scale). But the text says that my ancestors lived in the past 200 years in just one of these 4 highest-matching countries (not the right one anyway).
Also in a 1.4% Scandinavian category, 23andMe confidently picks Sweden as “where the ancestors lived in the past 200 years”. But in the details spreadsheet, one can see that the match to Sweden is as weak as it could possibly get (1 on a zero-to-5 scale). And it’s most definitely mistaken; no way anyone from Sweden could have moved to anywhere where the actual ancestors lived anytime after the year 1818
The new approach significantly adjusted the relative proportions of Japanese, Korean and Chinese estimated for my family members. I have one child who reads much ore Japanese than Korean, despite my other child and their mom showing the reverse. Some of that could be legitimate random chance, but a lot of the Japanese-Korean issue is methodology driven with a lot of Korean source ancestry in Japanese people from Yaoyi migration treated as Japanese instead.
Its still made from and for US Americans primarily. They dont have enough Central European diversity in their samples and what they have seems to be categorised more base on geography rather than ethnicity.
F.e. both me and my wife get significant British even though thats definitely wrong. I guess thats mostly because of the dominance of American samples than common Saxon heritage.
I have similar concerns to ohwilleke. I know a woman who is Korean and her 23andme ancestry is a bit suspect. It calls her as something like 40% Korean, 35% Chinese, 15% Japanese, etc. However, it calls her parents as 100% and 98% Korean.
Didn’t realize they did this update. Checked online. My 0.1% Native American has vanished, which isn’t entirely unexpected, but a bit disappointing, because my ancestry is (aside from the crypto-Sephardic g-g-grandfather) pretty boring.
I was hoping they did something to improve their French/German reference, but no such luck – I still show up as 30.8% British/Irish and 19.4% French/German on speculative, even though that’s impossible. My 100% German grandmother (who is still alive and had German-speaking parents born in Austria-Hungary) still comes out as only 31.2% French/German, FWIW.
The only additional information when I check the scientific details page is my British/Irish ancestry has a five-star match for Ireland, and only two for the UK. This is broadly correct, because I have a lot more Irish than English ancestry.
My ancestry results haven’t changed much but those of my s.o. have. She’s now showing as 71.6% British and Irish with the most recent ancestor being one generation back. She has no British or Irish ancestry.
She is, however, Breton, and I don’t know if 1100 to 1700 years (colonization of Brittany by British Celts occurred from the 3rd century on ) is enough time to be able to distinguish populations. I also suspect that there aren’t too many breton samples in the database since submitting dna for testing is technically illegal in France, even if it’s your own.
Razib,
Do you think it’s better to make a genetic test with LivingDNA or 23 andme or it doesn’t matter as you can put later your rawdata on second hand genetic analysis tools.
Otherwise, I submit you different questions I’d like to be discussed one day in your blog :
-How do people of european origins from USA, Canada, Australia and New Zealand fit among modern europeans ?
-Is there variation in old genetic admixture regarding socioeconomic status ? We know that genetic admixture vary between castes in South Asia but what do we know about it concerning Europeans (EEF, WHG, Yamnaya, …) or Japanese for example (yayoi / ainu) ?