Announcement

Collapse
No announcement yet.

Explanation of changes in My Origins v. 3

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Explanation of changes in My Origins v. 3

    There is much grumbling about the changes between My Origins v.2 and V.3. Many of us have felt rather uprooted by the changes. Could FamilyTreeDNA please explain the rationale that turns formerly British Isles origins into Central Europeans, and many other such changes. Many of us in the US trace our ancestral roots back to the countries from which our ancestors migrated to America, and the new version 3 seems to go back to a point where, for example, the Anglo Saxons were living in Germany. Where could I find more information?

  • #2
    I read at some other site (not sure where) that FTDNA will release a myOrigins Methodology Whitepaper about myOrigins 3.0, one would hope in the near future. This should be true, as they did it for myOrigins 2.0. I found a link to the whitepaper for myOrigins 1.0 in 2014 saved at the Web Archive (courtesy of Debbie Kennett's blog), and also a 2017 update (perhaps for myOrigins 2.0?) which you could check for an example of what to expect. The 2017 update does not include as much information as the 2014 version, but does show an increased list of reference populations. We can expect a myOrigins 3.0 Methodology Whitepaper to update the reference populations accordingly, and perhaps describe the rationale for the new population clusters.
    Last edited by KATM; 11 October 2020, 04:39 PM.

    Comment


    • #3
      In the email I got saying my updated information was ready, there was a link to a family tree dna blog which had an article talking about the new my origins 3, and in the section on reference populations it gave an example of lumping people in together if all four of their grandparents were born in the same place. I'm not sure if that indicates that FTDNA built their algorithm that way, but I certainly hope not because all four grandparents really doesn't get it: what if one of them was born to expatriate parents?

      Comment


      • #4
        Originally posted by Jeff H View Post
        There is much grumbling about the changes between My Origins v.2 and V.3. Many of us have felt rather uprooted by the changes. Could FamilyTreeDNA please explain the rationale that turns formerly British Isles origins into Central Europeans, and many other such changes. Many of us in the US trace our ancestral roots back to the countries from which our ancestors migrated to America, and the new version 3 seems to go back to a point where, for example, the Anglo Saxons were living in Germany. Where could I find more information?
        Along with those who grumble there are many who are very happy.

        Have your read all of the information here: https://learn.familytreedna.com/fami...tion-clusters/ and the FAQs here: https://learn.familytreedna.com/auto...ked-questions/

        Did you know "myOrigins results are your personal genetic ancestry that reflects the last 100 to 2,000 years (about four to 80 generations). They may also reflect one population that mixed with another in ancient times and became fixed in one of your populations." I don't know where my ancestors live 2,000 years ago. Do you?

        Comment


        • #5
          I'm not happy with the description of myOrigins (any version) as representing "your personal genetic ancestry that reflects the last 100 to 2,000 years (about four to 80 generations)". That's just bogus. What it represents is the degree to which your DNA is similar to that of various statistically-derived "reference groups" made up of modern, and probably almost entirely still living people who claim ancestry from a particular place and who are, within a reference group, more or less similar to each other. We have no way of knowing how long the ancestors of these reference people lived in any particular area, nor even if, for the most part, those ancestors actually lived in the same place 100 to 2,000 years ago. The only way we are going to know the genetic make-up of the population of, say, France 500 years ago, is to dig up a very large sample of datable bones, sequence their DNA, and also check for isotope markers to be sure they really are the bones of people who grew up in the same area where their bones were found.

          Comment


          • #6
            Let's add this article to the collection, to keep info in one place. The official ftdna blog:

            https://blog.familytreedna.com/myorigins-3-is-here/

            Comment


            • #7
              What I find bizarre is how significant ancestries disappear. In V2, I had 6% of sephardic jewish ancestry, which is consistent with the fact that I am descendent of a well-documented sephardic lineage. Then, in V3, this ancestry simply disappears, but I now get Sardinian (9%) and Peninsular Italian (7%), which I did not have before. My wife had some 25% of Southeast European ancestry in V2, consistent with the Italian surnames in her family. Now with V3, she has no Italian ancestry, but only some 4% of Greek...

              Comment


              • #8
                Emona, my belated thanks for the link. I'd seen the page previously, but hadn't looked it up again.

                I notice that the one mention in the FTDNA blog post about myOrigins 3.0, of "500 to 2000 years" is in this paragraph:
                At the subcontinental level, many population groups over the course of the past 500 to 2000 years experienced high mobility and interaction and because of that, the genetic differences are minor, so random recombination can play a role in whether a tester’s result matches the region their ancestors are from or a region neighboring it.
                also, further on:
                The random recombination process can sometimes cause some autosomal DNA to remix to look like a nearby reference population that is not seen in either parent or to appear in a greater percentage for a reference population than seen in either parent combined. This is completely normal.
                and
                Another reason this can vary so much is because it’s looking for the closest estimate to what your DNA matches. When new reference populations are added into the mixture or if you test with a new company that has completely different reference populations or is looking at a different data set of autosomal SNPs, you may end up with a different story of where your ancestors came from. While we’re looking for the closest match to your DNA, if your ancestors come from a population that is not represented, we will find your nearest DNA neighbor.
                I suppose those descriptions are true enough, and cover all possibilities for why the estimates may not reflect known ancestry. While new population clusters may clarify some customers' results, for other customers the new clusters instead muddy the waters. And so it goes with these estimates. If you check your match list, the origins of your matches is a more accurate reflection of your ancestry.

                Comment


                • #9
                  KATM, I agree those are appropriate caveats to keep in mind when attempting to square MyOrigins (or any other admixture) results with your own, known ancestry.

                  What hasn't been established, as far as I know, is the extent to which these (or any) results are realistic and don't have to be explained away by these excuses. A few years ago, Ancestry took a stab at verifying the results by measuring whether children's results were consistent with the results of both parents. If the algorithm is working as expected, children should generally not show up with a significant ethnic component that neither of their parents has. Ancestry did not actually report any numbers for this analysis, but they did indicate that their new algorithm performed better in this respect than their previous one. But when their next, latest and greatest algorithm was presented not too long ago, there was no mention of any statistics that might help to validate it.

                  Comment

                  Working...
                  X