Announcement

Collapse
No announcement yet.

TMRCA Calculations

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

  • TMRCA Calculations

    How are people estimating TMRCA from BigY results? Yes, I know it's fraught with uncertainty and maybe useless, but I'm just curious.

    So what is the process for looking at the total number of tested SNPs, novel SNPs and SNPs shared with matches to come up with an estimated GD and TMRCA?

    Is anyone using Doug McDonald's updated TMRCA calculator for this purpose? If so, what are the input parameters you are using; I couldn't figure out what to enter. Calculator is here: http://dna-project.clan-donald-usa.org/tmrca.htm

    Related question: When filtering BigY matches by terminal SNP, am I correct in assuming that those who have not tested to my terminal SNP on the FTDNA haplotree would appear in an upstream subclade? I ask because my FTDNA terminal SNP is not tested in BigY and I am trying to sort through my matches to determine which might be worth investigating.

    Thanks,

    Jim

  • #2
    Originally posted by jbarry6899 View Post
    How are people estimating TMRCA from BigY results? Yes, I know it's fraught with uncertainty and maybe useless, but I'm just curious.

    So what is the process for looking at the total number of tested SNPs, novel SNPs and SNPs shared with matches to come up with an estimated GD and TMRCA?

    Is anyone using Doug McDonald's updated TMRCA calculator for this purpose? If so, what are the input parameters you are using; I couldn't figure out what to enter. Calculator is here: http://dna-project.clan-donald-usa.org/tmrca.htm

    Related question: When filtering BigY matches by terminal SNP, am I correct in assuming that those who have not tested to my terminal SNP on the FTDNA haplotree would appear in an upstream subclade? I ask because my FTDNA terminal SNP is not tested in BigY and I am trying to sort through my matches to determine which might be worth investigating.

    Thanks,

    Jim
    Some of the Big-Y testers in the terminal SNP group that I belong to share 29 SNPs under Z156 and a couple of them have 10 singletons each. Dr. Iain McDonald has estimated a TMRCA for Z156 around 4,500 ybp so the testers in our group could divide 4,500 by 39 to get an average for their SNP line. In the coming Big-Y tests some men may have more or less SNPs than others. So, just like YSTRs some SNP Y-lines have more mutations than others. I hope this helps.

    Comment


    • #3
      Thanks,

      But I'm still confused about the process. The new McDonald TMRCA calculator appears to be able to use BigY data but I can't figure out how to enter the parameters. And Roberta Estes' blog makes a TMRCA estimate, but I can't determine how she did it. See: http://dna-explained.com/2014/06/24/big-y-matching/

      Comment


      • #4
        Originally posted by jbarry6899 View Post
        Thanks,

        But I'm still confused about the process. The new McDonald TMRCA calculator appears to be able to use BigY data but I can't figure out how to enter the parameters. And Roberta Estes' blog makes a TMRCA estimate, but I can't determine how she did it. See: http://dna-explained.com/2014/06/24/big-y-matching/
        I am not sure how the U106 Big-Y group got their estimates either.I think that all the 200 plus U106 Big-Y testers have an average of 36 variants below U106. That means that they all share the 1500-2000 YSNPs back to "Y-Adam" which they estimated to have lived 200,000 years ago.
        Armando may know more about the process. Have You asked any of the U152 project admins about the age of your group?

        Comment


        • #5
          Originally posted by 1798 View Post
          Have You asked any of the U152 project admins about the age of your group?
          Not yet--but I'm asking a different question. McDonald and Estes are doing direct TMRCA comparisons between two individuals. I am asking how they are doing that--what is the process and what are the inputs?

          Comment


          • #6
            Originally posted by jbarry6899 View Post
            How are people estimating TMRCA from BigY results?
            One must first apply criteria to distinguish reliable from unreliable SNPs. One consistent way to do this is to submit the BAM file to YFull for professional analysis.

            One can then consider "best-quality" and "acceptable-quality" SNPs to be reliable. Each such SNP appears to correspond to 140-150 years of patrilineage.

            So for example, if YFull determines that you have 11 novel SNPs (best-quality + acceptable-quality only), then you and the nearest other YFull customer share a MRCA roughly 1540-1650 years ago.

            Comment


            • #7
              Originally posted by lgmayka View Post
              So for example, if YFull determines that you have 11 novel SNPs (best-quality + acceptable-quality only), then you and the nearest other YFull customer share a MRCA roughly 1540-1650 years ago.
              Thanks--my YFull analysis should be done soon. In the meantime, I think I've figured out Doug McDonald's calculator. I read his description more carefully and the results seemed reasonable--roughly 52 generations (50% probability) using SNPs versus 40 using STRs. At least the weren't an order of magnitude different.

              Jim

              Comment


              • #8
                This is difficult. As far as my calculator, just enter (for BigY) 1.e7 (or 10000000) for Number of Markers and 3.e-8
                for mutation rate.

                The hard part, very hard, is getting the number of nonmatching markers.
                THIS CANNOT BE DONE USING JUST THE
                FTDNA WEB PAGE RESULTS!!!


                That's because in the "match" panels, if you try to
                use the number of unique markers present in a person,
                from their list, YOU CANNOT TELL IF THE OTHER PERSON
                HAS AN ANCESTRAL RESULT FOR ALL THOSE MARKERS OR NOT!

                This is because FTDNA is not taking into account nocalls.
                To get a correct answer you MUST download the .bed
                file and look up each and every marker in the match
                lists to see if they are called in both files. The mutation
                rates assume reliable markers only. So if one person
                is a no-call you neglect that marker.

                This process is merely painful. Its not hard. I have described how it do it several times.

                On the other hand, if you just add up the number of
                unshared unique variants for both people, a VERY
                rough guess is to divide that number by two. I have not tried to measure it accurately.

                I have tried several times to get FTDNA to mark which markers on the match list are no-calls in the "other person". And these attempts have all disappeared into a black hole. I tried to explain it and impress it on
                FTDNA people at the I4GG conference and apparently it
                did not register. Their inborn "simplicity is a commercial necessity even if its wrong" filter is too strong.

                Comment


                • #9
                  Thanks very much! Guess I have my work cut out for me this weekend!

                  Comment

                  Working...
                  X