Announcement

Collapse
No announcement yet.

Minimum

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

  • Minimum

    Is there any accepted calculation of the minimum number of siblings needed to test in order to fully reconstruct the genome of a parent? My best guess is 3, but that's just based on one observation of siblings whose phased kits at Gedmatch show ~2k shared paternal cM.

    Also a tangential thought: Is there a significant difference in the volume of shared cM for identical twins vs. any other type of siblings?

  • #2
    I don't think you'd ever be able to construct the full DNA of a parent. Usually some of their DNA will not have been passed to any of their children.

    n= number of children
    The average amount of DNA you'd be able to cover is
    100(1- 1/2^n)%

    e.g.
    1 child - 50% of their DNA, 100(1 - 1/2)%

    then on average:
    2 children - 100-25% = 75%, 100(1- 1/4)%
    3 children - 100-12.5% = 87.5%
    4 children - 100-6.25% = 93.75%
    5 children - 100-3.125% = 96.875%

    With each additional child you test you halve the amount of "unknown" DNA...on average.
    Last edited by ltd-jean-pull; 7th November 2017, 07:47 PM.

    Comment


    • #3
      Thanks.

      But I am only interested in the narrow range of their genome reflected in genealogical autosomal testing. So a bit of exaggerated imprecision in the original formulation of my question.

      My sense is that for all practical intents and purposes, 5 siblings is a very safe number since the estimated loss % is so small that it most likely--though not certainly--falls in the range of un-tested SNPs.

      Comment


      • #4
        I have recreated my Grandparents using my mother and 2 of her siblings (have a fourth but not added it yet) and my father and 3 of his siblings. All Family Finder Raw Data.

        I will figure out my average for both groups and let you know.

        Note, besides siblings you will need another close relative tested also if neither parent is tested to distinguish between siblings maternal vs paternal values per chromosome.

        Comment


        • #5
          That is awesome.

          You have all ~3,587 half-identical/~7,174 fully-identical cM that Gedmatch returns for an FF kit? From just 3 kits?

          Awesome.

          Comment


          • #6
            Grandparents are phased per chromosome into their Parent A vs Parent B chromosomes, but I have not identified which Great Grandparent is which as of yet.

            These kits are on Gedmatch
            Maternal Grandmother - To knowledge only english ancestry - T645924
            Maternal Grandfather - Norwegian Father, Swedish Mother - T837121
            Paternal Grandmother - Born in Romania, with German paternal Grandfather - T532230
            Paternal Grandfather - Scottish Father, Irish Mother. - T385374

            Comment


            • #7
              Originally posted by prairielad View Post
              Grandparents are phased per chromosome into their Parent A vs Parent B chromosomes, but I have not identified which Great Grandparent is which as of yet.

              These kits are on Gedmatch
              Maternal Grandmother - To knowledge only english ancestry - T645924
              Maternal Grandfather - Norwegian Father, Swedish Mother - T837121
              Paternal Grandmother - Born in Romania, with German paternal Grandfather - T532230
              Paternal Grandfather - Scottish Father, Irish Mother. - T385374
              Kit T628732 is my maternal Aunt that I did not use yet. Comparing her to my maternal Grandparents shows her not matching up as full parent child due to the DNA she inherited that the other siblings did not. This DNA is currently missing from my maternal Grandparents files.

              Comment


              • #8
                Originally posted by Frederator View Post
                Thanks.

                But I am only interested in the narrow range of their genome reflected in genealogical autosomal testing. So a bit of exaggerated imprecision in the original formulation of my question.

                My sense is that for all practical intents and purposes, 5 siblings is a very safe number since the estimated loss % is so small that it most likely--though not certainly--falls in the range of un-tested SNPs.
                I have also tested myself and 5 Siblings, between the six of us we do have 100% coverage of both parents

                Comment


                • #9
                  Thanks. This is a more thorough practical example than I could have hoped for.

                  But I'm not sure I understand some key points yet. How is it possible that you have phased your grandparents' kits into parent A and parent B if you haven't firmly established a kit/kits representing specific great grandparent(s) as a reference? I thought you needed a minimum of one parent (or a collection of close cousins from one side of the family) to phase a kit.

                  Sorry if those questions are phrased a little awkwardly. Hopefully you get my gist.

                  Comment


                  • #10
                    Originally posted by Frederator View Post
                    Thanks. This is a more thorough practical example than I could have hoped for.

                    But I'm not sure I understand some key points yet. How is it possible that you have phased your grandparents' kits into parent A and parent B if you haven't firmly established a kit/kits representing specific great grandparent(s) as a reference? I thought you needed a minimum of one parent (or a collection of close cousins from one side of the family) to phase a kit.

                    Sorry if those questions are phrased a little awkwardly. Hopefully you get my gist.
                    Where siblings share DNA they will be one of the following
                    1)Full base pair matching - Sharing DNA from the same paternal Grandparent on paternal Chromosomes, and Sharing the same maternal Grandparent on maternal chromosomes.
                    2)Half base pair matching - Sharing either the same maternal or paternal Grandparent, but receiving the opposite Grandparent on their other chromosomes. ie Sharing maternal Grandmother on maternal chromosomes, but one inherited paternal Grandmother (fathers maternal chromosome), and the other inherited paternal Grandfather (fathers paternal chromosome).
                    3)Not matching - each inherited the opposite maternal and paternal grandparents.

                    Thus during reconstruction/phasing, the parents chromosomes are phased per chromosome into their parent A and Parent B chromosomes (which in my case, since it is my parents and their siblings results, it is my Great Grandparents)

                    Edit
                    Results in picture are phased in better detail due to using 15 Family Members kits to determine individual values. (using just 3 siblings,the individual values would not be able to be phased to precision shown)
                    Attached Files
                    Last edited by prairielad; 7th November 2017, 09:47 PM.

                    Comment


                    • #11
                      I'll have to re-read this to be sure I really understand.

                      But my sense is that it is introducing some information that is new to me--suggesting that the testing process itself clearly distinguishes between the A & B chromosome copies. If true, I had long suspected it, but never knew that for an actual fact. Probably I should have known this earlier but I did not.

                      Also if that is true, then, once I have a 2nd sibling of my own tested, I will theoretically already have enough information to phase my father's (at least nearly) 100% reconstructed kit into partial kits for both of my paternal grandparents.

                      My paternal grandmother's matches are genealogically and ethnically distinct from my paternal grandfather's because she was from eastern Europe and he was from western Europe. If I designate one genealogical match from each chromosome as representing A or B, by default the other chromosome must represent the contribution from the other grandparent.

                      These would only contain 1/2 the cM of an ordinary living donor, but it would still be interesting and useful.

                      Even if that is technically correct, is it something the interface of the Gedmatch reconstruction utility could even handle? Or does their format require full siblings?

                      Comment


                      • #12
                        Originally posted by Frederator View Post
                        I'll have to re-read this to be sure I really understand.

                        But my sense is that it is introducing some information that is new to me--suggesting that the testing process itself clearly distinguishes between the A & B chromosome copies. If true, I had long suspected it, but never knew that for an actual fact. Probably I should have known this earlier but I did not.
                        ........
                        In a sense, yes testing process can distinguish between parent A and parent B, but only in little segments, as DNA is broken down into little segments to be read by chip and then it is "pieced back together" using a "template" to give raw data (multiple strands of DNA used, not just one single chromosome). Thus it can not say with certainty which "segment" belong together to make each parent A and parent B single chromosome per pair. ( testing process a little more complicated then that....)
                        Last edited by prairielad; 7th November 2017, 10:40 PM.

                        Comment


                        • #13
                          Darn. Well, thanks at least for setting me straight.

                          Comment


                          • #14
                            Here are my figures per chromosome using 3 and 4 siblings
                            Figures are based on tested positions (tested SNPs) inherited per chromosome.
                            Factors in 2 Values per position (maternal and paternal)
                            Attached Files
                            Last edited by prairielad; 8th November 2017, 12:10 AM.

                            Comment


                            • #15
                              Those are remarkably close to Jean's calculations.

                              Comment

                              Working...
                              X