Announcement

Collapse
No announcement yet.

Triplets don't have identical origins?

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

  • Triplets don't have identical origins?

    Anyone care to comment on the recent "Inside Edition" segment that featured identical triplets whose Family Finder tests didn't have identical percentages for their geographical origins?

    "How Accurate Are Home DNA Ancestry Kits? Investigation Uses Triplets to Put Companies to the Test"
    http://www.insideedition.com/investi...mpanies-to-the

    Thank you,

    David

  • #2
    It doesn't surprise me at all because I have long believed the ethnicity part of the DNA tests was unreliable. I believe it started as a fun extra, but Ancestry found out they could sell more tests by promoting it as the main reason for testing. Now I'm beginning to wonder if the cousin lists are reliable past 3rd cousin!

    Comment


    • #3
      One thing I would question is how their Raw Data matches up per tested SNP.

      This could possibly skew the admix results depending on how many variations in SNP there are as well as the fact that each would have different SNPs that have been entered as no calls due to the test failing to read that position/SNP

      These admix calculators only look at a certain set of SNPs to my knowledge, so depending if there are variation due to misread or mutations and no calls, I can see each having a slightly different admix result.

      Comment


      • #4
        From what I understand even one mutation can change the myOrigin algorithm from showing a good match to the Scandinavian cluster to either the British Isles, or Central European. I heard that the triplets were all 100% European and it was the sub continental results were different.

        Someone shared a few academic studies showing how the DNA in identical twins will not be exactly the same. I have linked them below (I thought they were interesting reads):

        http://www.fsigenetics.com/article/S...abstract?cc=y=
        http://www.pnas.org/content/102/30/1...urnalCode=pnas
        http://onlinelibrary.wiley.com/doi/1...1780360902/pdf
        https://www.sciencedaily.com/release...0328151740.htm

        -Darren
        Family Tree DNA

        Comment


        • #5
          Originally posted by DavidB View Post
          Anyone care to comment on the recent "Inside Edition" segment that featured identical triplets whose Family Finder tests didn't have identical percentages for their geographical origins?

          "How Accurate Are Home DNA Ancestry Kits? Investigation Uses Triplets to Put Companies to the Test"
          http://www.insideedition.com/investi...mpanies-to-the

          Thank you,

          David
          Thanks for bringing that up. I found the segment that actually aired on Youtube at https://youtu.be/qyfWZZ7uPuE The results start at 2:10.

          Unfortunately they didn't test both sets of identical triplets and the identical quadruplets at all three companies. If they had I think that 23andme would have come out the clear winner.
          Last edited by Armando; 25 February 2017, 10:39 PM.

          Comment


          • #6
            Some admixture tools use a random element for more efficient computation (e.g., analyze a random subset of say 10,000 SNPs, then another, and then another for some number of repetitions), so you can get somewhat different results even from the same data file. This has been reported by people who uploaded the same file multiple times to DNA.Land.

            Comment


            • #7
              A person with a lot of money to throw around could buy 3 different tests at each of the 3 companies and test himself or herself, using a separate account and a fake identity for each one.

              You would have 9 tests of the same person, 3 at each company. I bet the results would still be different for most of them.

              Comment


              • #8
                Originally posted by Darren View Post
                From what I understand even one mutation can change the myOrigin algorithm from showing a good match to the Scandinavian cluster to either the British Isles, or Central European. I heard that the triplets were all 100% European and it was the sub continental results were different.

                Someone shared a few academic studies showing how the DNA in identical twins will not be exactly the same. I have linked them below (I thought they were interesting reads):

                http://www.fsigenetics.com/article/S...abstract?cc=y=
                http://www.pnas.org/content/102/30/1...urnalCode=pnas
                http://onlinelibrary.wiley.com/doi/1...1780360902/pdf
                https://www.sciencedaily.com/release...0328151740.htm

                -Darren
                Family Tree DNA
                The first link showed that it took ultra-deep next generation sequencing at 91x to find just 5 SNPs that were different between twins and none of them were in dbSNP. That makes me doubt that the more limited testing by the three companies would have found SNPs that are actually different. Not that it isn't possible. Just like prairielad mentioned, the doctor in the show should have had someone check the raw data files of all of them in order to determine if there are differences are in the raw data and how much. I imagine that misreads or no-calls would be more likely than actual DNA differences to be the cause.

                Originally posted by Ann Turner View Post
                Some admixture tools use a random element for more efficient computation (e.g., analyze a random subset of say 10,000 SNPs, then another, and then another for some number of repetitions), so you can get somewhat different results even from the same data file. This has been reported by people who uploaded the same file multiple times to DNA.Land.
                But does the same thing happen with FTDNA, 23andme, and AncetryDNA?

                Originally posted by MoberlyDrake View Post
                A person with a lot of money to throw around could buy 3 different tests at each of the 3 companies and test himself or herself, using a separate account and a fake identity for each one.

                You would have 9 tests of the same person, 3 at each company. I bet the results would still be different for most of them.
                I have heard of at least one person not liking the result from myOrigins and getting a 2nd test and different results but I don't know if there was a difference in the raw data or not. I assume there was and it wasn't something going on with the calculator.

                Comment


                • #9
                  Instead of a SNP mutation, it could be a no call at just a tiny handful of loci. Each could have had a few & I think this would have been enough to throw off the algorithms for each.

                  Timothy Peterman

                  Comment


                  • #10
                    Armando and TE Peterman,

                    I have to agree with you. I would think a no call would be the more likely reason for the difference. It would be interesting to see what may change when the new version of myOrigins finally arrives.

                    -Darren

                    Comment

                    Working...
                    X