Announcement

Collapse
No announcement yet.

Is it possible to separate this out somehow?

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

  • Is it possible to separate this out somehow?

    Two relatives of mine (they're my 2C1R) - lets call them R (male) and M (female) are doubly related to each other. They're first and second cousins to each other.

    R's paternal grandfather Alfred was my mother's paternal grandfather's brother.

    M's paternal grandmother Kate was my mother's paternal grandfather's sister.

    In addition R and M's mothers are sisters. Over at Gedmatch we can see that R and M match along the entire X-chromosome. They both inherited exactly the same X chromosome from a maternal grandparent.

    We know therefore that the other X-chromosome that M inherited (which doesn't match R) has come from her grandmother Kate.

    I'm trying to work out how to use this information. I have gone through their lists of X-matches on Gedmatch and excluded those who match R from M's list, but I don't think it's as simple as this.
    Last edited by ltd-jean-pull; 26th May 2017, 11:03 PM.

  • #2
    Bumping the thread for any suggestions.

    Comment


    • #3
      Since R could only inherit X from mother (sister of M's mother), all matches shared with M on X will/can be looked at as being linked to the two sisters line.

      Any X match on M list that is not on R's will point to Kates line.
      But females due to having two X's have many false matches also, so depending on size of segment, many that show X matching that are not on R's list maybe just IBS (false positive)

      One way to work around this, since R and M show full matching on X (R being male only has one X, so it will be phased by default) is to create a research file for M by phasing her X with R's (remove the values which match R).

      Replace M's X with phased results and reupload file to Gedmatch as a research file.
      This will eliminate the majority of false matching on X

      Requires one to manipulate and work with the raw data files.

      Comment


      • #4
        Originally posted by ltd-jean-pull View Post
        In addition R and M's mothers are sisters. Over at Gedmatch we can see that R and M match along the entire X-chromosome. They both inherited exactly the same X chromosome from a maternal grandparent.

        We know therefore that the other X-chromosome that M inherited (which doesn't match R) has come from her grandmother Kate.

        I'm trying to work out how to use this information. I have gone through their lists of X-matches on Gedmatch and excluded those who match R from M's list, but I don't think it's as simple as this.
        I had to make a chart to visualize this, which I'm attaching (I hope I understood everything correctly).

        What you've said makes sense to me: that R & M got the same full X from their mothers, and that M's other X came from her father, who got it from Kate. Therefore, Kate's X is likely a recombination of her father's and mother's X chromosomes, but it's possible, although less likely, to be a full X from either.

        Since Kate's brothers did not pass down their mother's X, M's other X is a good clue for Kate's parents' ancestors on the X-inheritance pattern. Have you filled in an X-inheritance chart, which would follow Kate's parents' ancestors, to help you see connections with any matches?

        For the GEDmatch X-matches that are left on M's list (after you remove those shared with R):
        Do they share a large enough segment to be useful? I thought I had some statistics for male vs. female X segment size, but can't find it. In general, segments shared between female to female matches will be smaller than those shared between male to male matches, due to the female recombination.

        On the other hand, if your cousin M. has a recombined X from her grandmother Kate, part of it likely came from Kate's father, so the path for those segments would be Male (Kate's father) > Female (Kate) > Male (Kate's son) > Female (M.). The segments of the X from this alternating M>F>M>F pattern should have the least amount of recombination, compared to a F>F>F>F pattern (most recombination), and should be larger and maybe older.
        Attached Files

        Comment


        • #5
          Originally posted by prairielad View Post
          Since R could only inherit X from mother (sister of M's mother), all matches shared with M on X will/can be looked at as being linked to the two sisters line.

          Any X match on M list that is not on R's will point to Kates line.
          But females due to having two X's have many false matches also, so depending on size of segment, many that show X matching that are not on R's list maybe just IBS (false positive)

          One way to work around this, since R and M show full matching on X (R being male only has one X, so it will be phased by default) is to create a research file for M by phasing her X with R's (remove the values which match R).

          Replace M's X with phased results and reupload file to Gedmatch as a research file.
          This will eliminate the majority of false matching on X

          Requires one to manipulate and work with the raw data files.
          Thank you! They both tested with FtDNA so the raw data files will have the same format so this is doable.

          There definitely has to be numerous IBS on Margaret's X-matches because she has 2000 on Gedmatch with a length of 8.9cM or more, whereas R has 136 X-matches longer than 7cM.

          Comment


          • #6
            Thanks KATM. Your chart is correct.

            Yes, I have filled in an X-chart, but it would help if I could hone in on people who match Kate's X-chromosome as there are currently SO many matches.
            Last edited by ltd-jean-pull; 8th June 2017, 04:23 AM.

            Comment


            • #7
              The mention of GEDmatch and the long list of X matches prompts me to mention that the One-to-Many list of X matches for females on GEDmatch should NOT be taken at face value.

              My observation is that females will typically have 10,000 to 20,000 X matches on GEDmatch, of which only the top 2,000 are displayed, based on the statistics listed at the bottom of the One-to-Many report. Males, on the other hand, typically have 100-200 X matches on the same report. However, if you now use the X One-to-One comparison tool at GEDmatch on each of these matches, you will very likely find that the vast majority of the supposed female X matches on the One-to-Many report turn out to be spurious, they are not confirmed using the X One-to-One tool, while the nearly all of the male X matches (at least those over 10 cM) are confirmed as real. The discrepancy is so extreme, I think it is almost certain that there is a major flaw in the matching algorithm.

              Now if you replace the female kit with phased data, the zillions of spurious matches will disappear, and you will find around 100-200 X matches, most of them real.

              Remember, too, that female X matches should ideally be triangulated, and of course all the combinations within the triangulation group should be verified using the X One-to-One tool, so that you can be sure you are actually following the same segment from one generation to the next. This is necessary because females have two X chromosomes. For males, this is not an issue, because they only have one X chromosome.

              Once the X matches have been verified in this way, it MAY be possible to work out the connections, but there are often too many possibilities and not enough paper trail. Using X matches for genealogical problems isn't as easy as we might wish! A better X One-to-Many tool would help.

              Comment


              • #8
                Originally posted by John McCoy View Post
                Now if you replace the female kit with phased data, the zillions of spurious matches will disappear, and you will find around 100-200 X matches, most of them real.
                I've seen one female who had over 50,000 matches.

                With all respect John, how do you know that these 100-200 X matches are real after phasing? The problem is that virtually all of the X-matches are resulting from paths to a common ancestor from both parents coming together to reconstruct segments of this common ancestor meeting our criteria of a "valid" segment. Which side do you throw away with phasing? Occasionally after phasing you have enough DNA from the ancestor left to declare the residual as a valid segment, but the truth is being obfuscated.

                Since males do not have two X-chromsomes to use for reconstructing the common ancestor's DNA, you get sparse results.

                If we could reconstruct the common ancestor's X-chromosome, we would be able to filter out the matching segments that he is responsible for. The remaining small segments then could tell us something meaningful about other common ancestors.

                Jack Wyatt
                Last edited by georgian1950; 8th June 2017, 11:21 AM. Reason: syntax

                Comment


                • #9
                  By "real", I mean only that the One-to-Many X matches obtained for females using phased data can be confirmed using the X One-to-One tool. And I'm basing my observations only on segments of around 10 cM or longer. Where those segments came from is another question entirely, but using X matches for genealogy would be a lot more feasible if the GEDmatch One-to-Many algoroithm for the X chromosome were the same one used by the X One-to-One tool.

                  An interesting experiment, if anyone had the time, would be to check to see whether the "real" matches that are obtained when phased data are used (that is, confirmable by the X One-to-One tool), are the same ones as the few confirmable X matches found with unphased data for the same person. If the matching algorithm means anything, this ought to be the case, but I don't have the patience to wade through all of the One-to-Many X matches to find out.

                  Comment

                  Working...
                  X