Announcement

Collapse
No announcement yet.

False positives - now happened twice

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

  • False positives - now happened twice

    My father's just uploaded his Autosomal DNA to FTDNA. (Y - DNA kit being ordered).
    Aside from his match with me, he shares 134 cM with his top match - details below for >7cM lengths:

    Chromosome 3: Start 98151736; end 110445556; length 9.45 SNPS 1365
    Chromosome 17: Start 6433922; end 10856906; length 12.53 SNPS 1021
    Chromosome 18: Start 64307610; end 75328906; length 31.04; SNPS 2374
    Chromosome 19: Start 211912; end 2986766; length 7.93; SNPS 589
    Chromosome 19: Start 4049947; end 37211288; length 35.16; SNPS 3205

    I believe this is a false positive (identical by state) not a 2nd/3rd cousin.

    This is because I posted a query last week - where I do not match a person on FTDNA who is my top GEDmatch match. Thanks to help from this forum, it was concluded that my top GEDmatch "match" was highly suspect - for reasons including long stretches of DNA on the higher chromosomes and matches at the start of chromosomes.

    It now turns out that my top GEDmatch "match", who I do not match on FTDNA, is the very same person as my father's top FTDNA match (set out in the table above)!
    Again, I can see the same pattern of long stretches of DNA on the higher chromosomes and at the start of chromosomes.

    Could someone be kind and explain why these false positives of "matches" on the higher chromosomes/start of chromosomes are happening? Is it just random?
    I've tried googling but I'm not sure I'm using the right terminology, so perhaps there's some reading that people could recommend?

    Thanks

    Liz

  • #2
    I am extremely dubious about the supposed false matches involving "higher [numbered] chromosomes and at the start of chromosomes". This seems especially doubtful here, given the long segment cM lengths and the high numbers of SNP's for each segment. What is the evidence for these phenomena?

    I would be more inclined to accept what GEDmatch says, because there appears to be no behind-the-scenes editing of matches on that site. If it were me, I would be looking only at segments around 10 cM or greater, and I would be looking at all combinations of the kits involved to see if there were any logical inconsistencies.

    Comment


    • #3
      Originally posted by John McCoy View Post
      I am extremely dubious about the supposed false matches involving "higher [numbered] chromosomes and at the start of chromosomes". This seems especially doubtful here, given the long segment cM lengths and the high numbers of SNP's for each segment. What is the evidence for these phenomena?
      I strongly agree. With that many segments and especially two of the segments 30+ cM (chromosomes 18 and 19), these are not "false positives." If I had to guess, I would say that the father of the OP and this match are probably something like 2nd cousins, once removed.

      Comment


      • #4
        Based on the new evidence of your father, I will say that I now believe this is a true match (and not a false positive).

        I noticed you have started the loading of your father to Gedmatch which verifies fairly closely what your father's FTDNA numbers are saying. Of the matching segments, only the chromosome 22 is showing on Gedmatch, but not FTDNA.

        You did not mention how FTDNA classifies the rank to the match, but based on you saying it is a total of 134 cM, we can use the following table:

        https://isogg.org/wiki/Autosomal_DNA_statistics

        to show the ranking should be closest to the 106 benchmark of about a 2nd cousin once removed.

        Using the Match's gedmatch list, it shows the following matching:
        To your father: total 109 largest 40
        To you: total 50 largest 18

        Although total cM count on your father is showing a noticeable difference in how Gedmatch calculates (versus FTDNA), it still fits in the same class target of the 106 benchmark. The drop in DNA numbers to you is also consistent with you being one generation lower (in the tables, the benchmark would be 53.

        Overall, I would have to say you and your match should be 2nd cousins twice removed (or equivalent).

        Returning to the original question in your other thread of why FTDNA did not recognize this, I do not know. Perhaps the programming does need tweaking. If someone from FTDNA is reading this, it might be worth looking into.

        You may still not be inclined to do full autosomal testing, but I would think that full testing would certainly be enough to pick up your match (even if the match is ranked a class below where you would expect it).

        Comment


        • #5
          Liz,
          I do want to add a little practical advice, as the issues are now starting to diverge. My interest in this thread was based on the original post (on the other thread) about the differences between FTDNA and Gedmatch.

          If your primary interest is following your paternal line, then you what you want to do is to log to your father's FTDNA and check the match box of the high Match (not you, but the next one). Then you click the "in common" tab and it will give you a list of shared matches between your father and the match. You can then use the chromosome browser to see if the matches seem to be sharing the same segments. (You could also download the chromosome CSV file and view it in a spreadsheet). Using this triangulation together with what you find from the paper trail should bring you closer to what you want to discover about the convergence. If this approach fails, then you can consider doing a full autosomal test on your father to find the elusive missing link people that can tie it together.

          What I am trying to say is that for your ultimate goal, improved testing of your father is more beneficial than additional testing on yourself. You can also think about the Y-DNA, but that of course will involve different matching methods.

          Comment


          • #6
            Thanks everyone for explaining this to me. I now understand far better. Luckily, I have a relatively full family tree, as does the cousin who matches my dad and I. But try as we may we can't, so far, find a match on a 2nd cousins once removed relationship. So two further questions from me - if anyone can help

            (1) Cousins
            The link to isogg.org/wiki/Autosomal_DNA_statistics from ech124 is really useful. It includes some alternatives, based on shared cMs, to second cousins once removed.
            I've ruled out the "first cousin three times removed" option due to the generation dates.
            This leaves "half first cousins twice removed" and "half second cousins".
            But I can't get my head around what a half cousin is ...... i.e. aside from matching on both a cousin's parents, aren't all cousins a match just on one of their parents? If so, how does this differ from a half cousin?

            (2) Share segments 2-6cM
            Looking at the 'Ranges of shared cM' on - isogg.org/wiki/Identical_by_descent#Thresholds_for_matches - the 3rd cousins range is between 16-111 cMs. My dad and his match share 106cM which fall in the range, albeit well above the expected share.

            There's also a column headed 'Range of number of shared segments' and for 2nd cousins once removed, this is listed as between 4-13 shared segments. For 3rd cousins, this is listed as between 2-6 shared segments.
            But shared segments over what cM?
            e.g. My dad and his match share 5 segments above 7cM and 8 segments over 3cM. So for the purpose of 'Range of number of shared segments' should I only be counting cM above 7, or not?

            Liz

            PS: I'm also trying to narrow matters down via the smallish, I think, shared X chromosome. But I'm posting this query the FTDNA X chromosome board, to keep the subjects separate.


            PPS: Thanks so much ech124 for the additional advice that goes well beyond my FTDNA/GEDmatch query. I've done as you suggested, used the 'in common with', found 7 people in common with my dad and his primary match, used the chromosome browser .... and it looks a shambles, with only two overlaps - on ch 3. Only one of these has a tree and no one in this matches my dad's name or is even in the right country (Wales).

            So a full autosomal test, for my dad, seems the way to go.

            My primary concern was simply breaking down a major brick wall on my direct paternal line. But I have other brick walls too. Many could be broken down if only I could just get back beyond the opaque window of 'no the paper records between 1780 - 1839' in my locality, due to Welsh non conformism. Prior to thus, the paper records become available again.

            Comment


            • #7
              The match may be just slightly more distant than you assume or even slightly closer, but not much, given the number and size of the segments.

              The other possibility is that either you or your cousin has an error in your tree. This can be due to faulty research, but it can also be due to misattributed parentage, often referred to as a non-paternal event. This happened to me. I had made a couple of trips to the courthouse and had proof of my great-grandfathers parents. But a Y-DNA test proved my documents wrong. My great-grandfather was not the son of his supposed father (his mother's husband), but the son of a neighbor. The court records didn't indicate that. There could also have been an informal adoption, etc.

              Comment


              • #8
                I can try to answer your question on half cousins.

                Two full siblings share both parents. When siblings only share one parent (most often because of a re-marriage) then they are half siblings.

                The children of the full siblings are normal full 1st cousins. The children of the half siblings would be half 1st cousins.

                The term "half" just keeps following down the generations. The distinction is important because the starting point of half siblings is at lower amount of shared DNA.

                I think you already understand the concept of removed as a difference due to generation.
                Last edited by ech124; 3 August 2017, 12:33 AM.

                Comment

                Working...
                X