Announcement

Collapse
No announcement yet.

FTDNA released their Chromosome Painting today

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

  • FTDNA released their Chromosome Painting today

    Roberta Estes has published a blog post today, "FamilyTreeDNA’s Chromosome Painting Just Arrived!!!," with links to three videos by FTDNA's Population Geneticist, Paul Maier, Ph.D. He discusses MyOrigins 3.0 in all three videos, including some coverage about the Chromosome painting feature released today. I just viewed these three videos.

    In the third video, he explains why it is hard to accurately assign the smaller continental regions (such as countries) to chromosome segments, so that is why FTDNA's Chromosome Painting is only to the "Super Population," or larger region, level. Thus, if someone has ancestry from three large sub-regions of a continent, they will show three colors in the painting (more accurately, three shades of one color). Someone who has ancestry from only one major region of a continent will only show one color. Those people who have ancestry from various regions of multiple continents will get the most "colorful" Chromosome Paintings. The X chromosome is not included in the Chromosome Painting, but autosomes 1-22 are included.

    Although he makes the case for why only the Super Population level is used for FTDNA's Chromosome Painting, one must wonder how 23andMe can include more granular detail in populations, such as Italian, British & Irish, French & German, North African, etc., whereas FTDNA can't or won't.

    In any case, it is worthwhile to view these three videos, as it explains the MyOrigins methodology. While the videos are fairly understandable, some of the terms are still too technical for the average person, though, in my humble opinion.

  • #2
    Originally posted by KATM View Post
    In any case, it is worthwhile to view these three videos, as it explains the MyOrigins methodology. While the videos are fairly understandable, some of the terms are still too technical for the average person, though, in my humble opinion.
    I also view the three videos yesterday. I agree with both points.

    Comment


    • #3
      Honestly a lame update... They should've supplied painting to sub-regions too as was advertised in their 'videos'

      I'm begging FTDNA to change this...Anyone else here willing to sign my petition? lol

      Comment


      • #4
        I'm underwhelmed. Kit turned out virtually all Western Europe, no further subdivision in spite of actual English, Scottish, and Irish ancestors circa 1800, plus less than 2 percent Southern European. The segments marked Southern European seem to coincide with triangulation groups from the branch of the pedigree that includes the one Jewish ancestor born in 1833 from the area around Szeczin, now in Poland. However, it is probably a good thing that FTDNA has chosen to be careful to avoid pushing the data too far. Previous attempts at chromosome painting (tools hosted on GEDmatch, for example) have been way, way too optimistic about assigning segments to all sorts of diverse "ethnicity" for the same kit, in a way that proved to have absolutely no predictive value, even though it was colorful.

        Comment


        • #5
          Originally posted by John McCoy View Post
          I'm underwhelmed. Kit turned out virtually all Western Europe, no further subdivision in spite of actual English, Scottish, and Irish ancestors circa 1800, plus less than 2 percent Southern European. The segments marked Southern European seem to coincide with triangulation groups from the branch of the pedigree that includes the one Jewish ancestor born in 1833 from the area around Szeczin, now in Poland. However, it is probably a good thing that FTDNA has chosen to be careful to avoid pushing the data too far. Previous attempts at chromosome painting (tools hosted on GEDmatch, for example) have been way, way too optimistic about assigning segments to all sorts of diverse "ethnicity" for the same kit, in a way that proved to have absolutely no predictive value, even though it was colorful.
          Pretty much the same here, I’m western european deep blue all over dotted with lots of turquoise eastern european, and it feels like there’s no rhyme or reason for any of it. I thought I remembered them saying you would be able to compare matches in chromosome painter but I guess I was wrong, so it kind of feels pointless. Still, it’s cool that they’re still making improvements.

          Comment


          • #6
            Rhonda Hatton, Jonny Perl posted yesterday in his DNA Painter Facebook Group that he is working on a tool to import the segment information from FTDNA's Chromosome Painting, but it isn't ready yet. But he also said that he made a quick release so that users could use the regular "import segment data" tool and select the FTDNA segments [from the?] CSV, and DNA Painter will recognize and import it.

            Comment


            • #7
              Chromosome Painting is assigning me segments on various chromosomes that don't show up in either of my parents. We all tested with FTDNA, so comparability of raw DNA data shouldn't be the reason. It doesn't make sense. When they update MyOrigins in the future, I presume this will apply to chromosome painting too? Hopfully the results will improve...

              Comment


              • #8
                Originally posted by StefG View Post
                Chromosome Painting is assigning me segments on various chromosomes that don't show up in either of my parents. We all tested with FTDNA, so comparability of raw DNA data shouldn't be the reason. It doesn't make sense. When they update MyOrigins in the future, I presume this will apply to chromosome painting too? Hopfully the results will improve...
                That's a big problem in the methodology. It remains to be seen how may people (or "trios", as the FTDNA "white paper" calls them) find themselves in this situation. The "trios" who find inconsistencies will post here, but we will probably never hear from the ones who don't, where the methodology did not produce this sort of artifact. Nevertheless, "consistency" is one of the few testable predictions of the chromosome painting statistical methodology. The degree to which parent-child inconsistencies are found is a measure of how often the methodology is failing to detect actual sequences that predict admixture, and, apparently, detecting false sequences instead.

                Comment


                • #9
                  My CP is okay although the MyOrigins it is based upon is pretty shaky.

                  But my mother's has several issues....


                  Chr 20, 21, 22 are completely duplicated. The Paintings are the same and the numerical data for each segment is just repeated within the pairs of chromosomes.

                  Capture20to22.PNG

                  ~~~~~~

                  And then there is also this issue. This segment is from my mother's results

                  Americas 20 ...2 .... 63,244 .... 4,797,409 .... 15.29


                  I seem to have inherited the same segment exactly. Same Start, Stop, cM.

                  Western Europe 20... 2....63,244.... 4,797,409 ....15.29

                  For my mother, its Amerindian. For me, its Western Europe. Hmmm....not exactly the same.


                  I have submitted a ticket to Customer Support. The original help person tried to tell me this is normal, but agreed to send it on to someone more familiar with the CP.

                  Comment


                  • #10
                    Finding exactly the same segments on the two homologues of three chromosome pairs sounds extremely unlikely. It implies recombination (a process that is mostly random) has occurred at exactly the same spots on both sides of your pedigree!

                    Yes, it IS possible that the segment labeled Americas for your mother and Western Europe for you is "real", in the sense that the second analysis is picking up a Western Europe segment with the same boundaries that came from your father, while the Americas segment from your mother has been disrupted by recombination and so doesn't register separately in your kit. Except that such a result requires the "duplicated" paintings in your mother's kit to be wrong. We would really like to know what "sequence" the methodology is inferring in each case. Is there a constraint in the methodoloogy such that when an inferred sequence is "informative" for admixture, that only the complementary sequence is used to detect admixture origins on the other, homologous chromosome? If that constraint is not in place, it would be possible for the methodology to detect two different origins at any location based on many of the same SNP values! We would also like to see actual triangulating matches (using the normal autosomal segment matching algorithm) with other kits for those segments, to verify that the sequences the chromosome painting algorthm has inferred are in fact that ones that appear to be "real", because they are shared with our relatives.

                    An interesting train of thought, what would happen if we restricted chromosome painting to those segments (sequences) that can be verified by strong matches with other kits? In other words, apply the chromosome painting algorithm only to those inferred sequences that appear among our matches. A new way to evaluate the performance of the chromosome painting methodology!

                    Right now, we don't know what those autosomal segment matching sequences are, for the most part, and that's another aspect of the genetic genealogy puzzle that needs to be visible. For any matching segment, we should be able to see what the sequence is. That particular issue will be forced, I think, as "long read sequencing" becomes more accessible, and we begin to compare segment matches based on inferred sequences derived from individual SNP scores with real sequence data. Yes, that will allow us to see matching segment sequences and hence some genotype data for individuals we match. I know some people believe that's an unacceptable breach of privacy. That issue will have to be confronted as well. The technology will move ahead regardless. I would much prefer real data to fake data.

                    Comment


                    • #11
                      I have examples of the same overlapping issue at 23andMe. I find a unique ancestry being doubled and overlapping on both homologues of the same chromosome. Although I think that is a different issue from FTDNA where the exact Start and Stop points for every segment are also being doubled.

                      This person has one Italian parent. But the Italian shows on both chromosomes. His other parent is British/Irish. This happens on several chromosomes but to a lesser degree.

                      .Capture23 Italian both chromosomes.PNG

                      Comment

                      Working...
                      X