Wednesday, 7 February 2018

Third-party admixture: DNA.Land, Gencove, GenePlaza

In this blog post, I will be presenting my admixture results from three popular third-party analysis providers: DNA.Land, Gencove and GenePlaza. These are websites where you can upload your autosomal raw data from your testing company and get a "second opinion" (so to speak) on your biogeographical ancestry, based on datasets and algorithms different from those of the testing companies. If you are considering uploading to any of these third-party sites, please be advised that you are doing so at your own risk. I do not explicitly endorse, and I am not affiliated with, any company or website.

I will discuss my third-party results with reference to both my known paper-trail ancestry as well as my admixture results from 23andMe, Family Tree DNA and MyHeritage. Please see my previous blog post for a detailed discussion about my original testing-company results.

DNA.Land

DNA.Land is a research project run by Columbia University and the New York Genome Centre. Uploading your raw data there means consenting to participating in research. The admixture and matching tools are a courtesy to the participant, and not the main purpose of the site.

Below is a screenshot of my DNA.Land results based on my autosomal raw data from FTDNA. The categories are very broad, with no separate Scandinavian category (most of it is probably incorporated into the "North/central European" category). Apart from that, this looks like a plausible breakdown for me.



This second screenshot is of my DNA.Land results based on my 23andMe raw data. As you can see, it is a little bit different from my FTDNA one, but the main categories and proportions are quite similar:


At first glance, the Balkan and Southwestern European percentages seem very interesting, since 23andMe does give me 0.5% Broadly Southern European. Perhaps this really is from Balkan and Southwestern Europe? The Kalash percentage is also very intriguing. However, none of my parents get any Kalash or Balkan or Southwestern European, so all of these percentages might conceivably be statistical noise.

My parents' results:

My father:


My mother:


Both my raw data files as well as my parent's files show up with Finnish DNA, which is good, but the percentage is very high, and some of it might in fact be Scandinavian DNA which is misread because there is no separate Scandinavian category.

Gencove


Gencove has a quite straightforward admixture analysis program with a visually pleasing map layout reminiscent of AncestryDNA, MyHeritageDNA and FTDNA myOrigins.

My European categories seem, in short, correct. There is not much more to say. Everything is roughly consistent with my paper trail as well as my results from the testing companies. Interestingly, the "Northern and Central Europe" category includes a large portion of the British Isles as well as the northwestern part of continental Europe. In my opinion, this seems like a very good representation of ancient migrations and gene flows across the English Channel.

Also, well done Gencove for reporting my Finnish DNA!

The Central Asian is a bit of a puzzle, just like the Kalash at DNA.Land; I will be discussing it at the end of the blog post.

GenePlaza

Ancestry
GenePlaza's Ancestry app has no specific Scandinavian category. The region of Scandinavia seems to be split between the "North Slavic" and "Northwest European" categories. The Southern European percentages (5.6% Southwestern European and 4.5% Eastern Mediterranean) are intriguing, and like the SW European and Balkan at DNA.Land, these categories might possibly represent real Southern European ancestors within a genealogical timeframe (cf. my 0.5% Broadly Southern European at 23andMe). GenePlaza also gives me a mysterious Central Asian percentage just like Gencove does; again, I will discuss this at the end of the post. The Central Asian category has no subcategories.

Interestingly, GenePlaza gives me 0.7% Ambiguous. I always like it when there is an Ambiguous or Unassigned category, because it means that the algorithm isn't force-fitting your genes into categories where they don't belong.


K12 Ancient Admixture Calculator
Besides the regular admixture analysis app, GenePlaza also has one that is specifically geared at showing you your ancient roots through comparisons with actual archaeological remains. This one is called the K12 Ancient Admixture Calculator. The results are not really relevant to genealogical research within a historical timeframe, but I still find them interesting, and I might do a separate blog post in the future about "ancient calculators" like the K12 and FTDNA's "AncientOrigins".


Conclusion

On a broad continental level, all these third-party sites are doing a decent job at predicting my admixture.

On a regional level, all the sites report my strongest European categories as being in the North, which is consistent with my paper trail and testing-company results. Several of the third-party sites find more Eastern/Northeastern European DNA ("North Slavic" etc.) than the testing companies do. I wonder if that could have something to do with my Finnish/Russian ancestry, or if it's just something generic found in all Scandinavians. Both DNA.Land and GenePlaza give me Southern European percentages, which could be real; however, if they are, the percentages are probably too high. 23andMe's 0.5% Broadly Southern European is more consistent with my known paper trail and with the fact that both my parents show 0% Southern European at DNA.Land.

Additionally, all the three third-party sites assign me some Central Asian under various labels. I don't know where this comes from, but the consistency indicates that it is probably a reflection of something real. Perhaps it could have something to do with my mother's alleged Russian ancestry? Our Russian ancestor might not have been 100% ethnically Russian; there might well have been some other admixture there, such as Tatar. The Central Asian DNA could also be connected to our Romani ancestry, or to our Finnish and Sámi lines.

In my opinion, out of these three sites, it is Gencove that provides the best admixture breakdown for me, while DNA.Land has the most coarse-grained and uninformative analysis. The fact that DNA.Land assigns me several categories at ≥4% while both my parents get 0% of the same categories, makes me question the reliability of DNA.Land's admixture breakdown more generally. The margin of error seems very high. However, DNA.Land has updated their admixture analysis before, and will hopefully do so again.

4 comments:

  1. Thanks for the tip about these sites. I uploaded my DNA to both Gencove and GenePlaza. Gencove's results were: 50% Northern and Central Europe; 24% Finland; 18% Scandinavia; 7% Northern British Isles; 1% Northern Italy. GenePlaza was: 61.6% Northwest European; 30.2% North Slavic; 6.5% Southwestern Europe (which includes Basque/French and Basque/Spanish) and 1.7% ambiguous.

    When I use the admixture calculators on GEDmatch, I always get a lot of Russian, Belorussian, Lithuanian and Ukrainian, which is born out on GenePlaza's North Slavic. Also I get a lot of Basque (both French and Spanish) on the calculators there. Pretty fascinating stuff. Thanks again!

    ReplyDelete
  2. You're welcome! Your results are interesting! It's cool that Gencove are splitting Italy up into different regions.

    Do you think your Basque is real or noise?

    ReplyDelete
  3. I have to think that the Basque is real. It comes up so much in the GEDmatch calculators, sometimes as the first thing in the Oracle 4 populations approximation. In the MDLP World22, for example, Basque_derived is in the first position in each of the 20 populations listed.

    AncestryDNA has me at 18% Irish and I found this interesting tidbit about a possible Basque/Irish connection:

    Are the Basques the Closest Genetic Relatives of the Irish?

    Today, people living the north of Spain in the region known as the Basque Country share many DNA traits with the Irish. However, the Irish also share their DNA to a large extent with the people of Britain, especially the Scottish and Welsh.

    DNA testing through the male Y chromosome has shown that Irish males have the highest incidence of the haplogroup 1 (or Rb1) gene in Europe. While other parts of Europe have integrated continuous waves of new settlers from Asia, Ireland's remote geographical position has meant that the Irish gene-pool has been less susceptible to change. The same genes have been passed down from parents to children for thousands of years. The other region with very high levels of this male chromosome is the Basque region.

    This is mirrored in genetic studies which have compared DNA analysis with Irish surnames. Many surnames in Irish are Gaelic surnames, suggesting that the holder of the surname is a descendant of people who lived in Ireland long before the English conquests of the Middle Ages. Men with Gaelic surnames, showed the highest incidences of Haplogroup 1 (or Rb1) gene. This means that those Irish whose ancestors pre-date English conquest of the island are descendants (in the male line) of people who probably migrated west across Europe, as far as Ireland in the north and Spain in the south.

    Some scholars even argue that the Iberian peninsula (modern-day Spain and Portugal) was once heavily populated by Celtiberians who spoke at now-extinct Celtic language. They believe some of these people moved northwards along the Atlantic coast bringing Celtic language and culture to Ireland and Britain, as well as France. Although the evidence in not conclusive, the findings on the similarities between Irish and Iberian DNA provides some support for this theory.




    ReplyDelete
  4. As a follow-up to my comment above, today I came across the FTDNA project for my father's last name. Within the Y-DNA results I found one of his ancestors who was listed as having the haplogroup R-M269 (from the Rb1 haplogroup mentioned above.) That haplogroup is found in high levels within the Basques. My dad's paternal ancestors were Mennonites who lived in various places within Switzerland, Germany, and France due to religious persecution. I don't think it would be too much of a stretch to posit that their ancient ancestors might have migrated from Spain and Portugal after the last ice age retreated. Maybe that is why Basque/French and Basque/Spanish shows up so much for me. Add that to the Irish/Basque connection and it makes sense. At least to me. :)

    ReplyDelete