Announcement

Collapse
No announcement yet.

How many mutations are too many?

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

  • How many mutations are too many?

    Hi

    My first post here. I've been the admin for a one name project [RID(E)OUT] for some time now; my members ask questions from time to time and I do my best to help them but I'm stumped with this one... can you help please?

    I have three members who, on paper, descend from three different sons of 'John' (b. 1745)... the men are five or six generations down from 'John' - two have identical sequences (37 markers), one has 1 fast mutation [DYS458].

    Now, a fourth man has joined the project; on paper, he believes himself to descend from a fourth son of 'John' yet, in just seven generations to the CMA, his line shows 3 mutations [DYS460, DYS464c, DYS576] all of which are, I believe, 'fast' - but is this number feasible over so few generations? Is it more likely that the man actually belongs elsewhere on the tree - or is this rate of mutation ever observed?

    Cheers, Karen

  • #2
    The difficulty is that there is no way to be sure that the fourth man was simply very lucky and that unlikely mutations actually happened in his lineage. What is needed here is a sense of exactly how unlikely this outcome is.

    Another way to proceed would be Big Y or a similar SNP-based analysis. If the first three samples do in fact belong on the same twig of the haplotree, as determined by careful analysis of Big Y or other SNP analysis, and if the fourth sample doesn't belong on that twig or one of its twiglets, then you would have evidence pointing to a different origin. Or, if it does cluster with the others, then you have evidence that he has a similar origin, not necessarily with the common ancestor shown by the paper trail, but at least from a close patrilineal relative.

    Also, the statistical context might be more persuasive if it turns out that the surname has many different, genetically distinct origins. If, on the other hand, the genetic evidence points to most or all such families having genetically similar origins, it becomes much harder to argue that one of them is different from the others.

    I have a feeling the genetic evidence may be able to rule out some possibilities, but it may not be powerful enough to distinguish ancestral brothers from other patrilineal relatives.

    Comment


    • #3
      How many mutations are too many?

      Hi John

      Thank you for your very prompt and helpful response.

      Your first sentence encapsulates my problem... 'what is needed here is a sense of exactly how unlikely this outcome is' and that is what I don't have. I have to admit that I do find that the statistical elements of some of these analyses a bit baffling; one can find mutation rates for individual markers but not the likelihood of three mutations occurring within one time period (7 gens, ~170yrs). I haven't looked into Big Y so I'll need to do some reading but, all supposing that it involves further testing, I'm not sure that all four of my members would be interested, particularly since three of them are comfortable that they know their lineage. Still, I will do a bit of research: I'm very grateful for your input so thank you again

      Cheers, Karen

      Comment


      • #4
        The most basic approach is to consider the 3 mutations as independent events, and compute the probability that all three sites mutated within the known number of generations. However, there are some cases where a whole stretch of the Y chromosome, containing several nearby STR's, has apparently undergone some sort of major reorganization, all at once. In other words, mutations of STR's may not always be independent events. Another complication that can't easily be measured is that there is no reason to suppose that every Y chromosome experiences the same mutation rate. A lot of unknowns!

        Comment


        • #5
          Originally posted by Bingowings View Post
          Your first sentence encapsulates my problem... 'what is needed here is a sense of exactly how unlikely this outcome is' and that is what I don't have.
          For an example of a large GD in a few generations see Prairielad's post where he says..
          Originally posted by prairielad View Post
          My father and his paternal 1st cousin are 4GD at the 37 and 67 marker level.

          My Uncle and same cousin are 3GD @ 37 markers

          My father and his brother are 1 GD @ 37 markers. Difference is my father registers a 2 step at DYS576 and my Uncle has only a 1 step.
          Originally posted by Bingowings View Post
          I haven't looked into Big Y so I'll need to do some reading but, all supposing that it involves further testing, I'm not sure that all four of my members would be interested, particularly since three of them are comfortable that they know their lineage. Still, I will do a bit of research:
          The BigY does involve further testing and it is expensive but it tests SNPs which can be much more reliable and SNPs can be dated since new SNPs happen about every 3-4 generations. You can see video where STR matches and BigY tests were used on a lot of people that matches at 67 markers at https://youtu.be/pxexkvfus6w proving that BigY helps determine how closely people are related to within about 100 years. It takes a $49 YFull analysis to get a TMRCA of the SNPs so you have to tack on that price to the BigY testing to understand how expensive it all is.

          Comment


          • #6
            Originally posted by Bingowings View Post
            Hi

            My first post here. I've been the admin for a one name project [RID(E)OUT] for some time now; my members ask questions from time to time and I do my best to help them but I'm stumped with this one... can you help please?

            I have three members who, on paper, descend from three different sons of 'John' (b. 1745)... the men are five or six generations down from 'John' - two have identical sequences (37 markers), one has 1 fast mutation [DYS458].

            Now, a fourth man has joined the project; on paper, he believes himself to descend from a fourth son of 'John' yet, in just seven generations to the CMA, his line shows 3 mutations [DYS460, DYS464c, DYS576] all of which are, I believe, 'fast' - but is this number feasible over so few generations? Is it more likely that the man actually belongs elsewhere on the tree - or is this rate of mutation ever observed?

            Cheers, Karen
            Do you have a public website for the project? If you do and if you'll provide the kit numbers for the four people we can view the results and perhaps make better suggestions.

            Comment


            • #7
              How many mutations are too many?

              Hi... in answer to your question, no we do not have a public website. However...

              Males 3, 4, 5 & 6 have 37 markers identical and an established CMA 6-7 generations back. Male 1 also has documentary evidence of belonging to this family and shares a closer relative with two others and so he has a single mutation (DYS458) in his own line. The chap (Male 2) who believes himself part of this group, on paper, is 7 generations down from the CMA but has 3 mutations as shown above. My original question was: is he likely to be related to this group or not based on the DNA results alone? The situation is 'difficult' as he is adopted but is in touch with his birth father, who passed on the genealogical information to him, much of which has documentary evidence.

              Thanks for looking at this; I would be very glad to have some opinions :-)

              Regards
              Karen
              Attached Files

              Comment


              • #8
                As Armando has shown with examples, a GD of 3 at Y37 can still mean they are from the same lineage. I have seen that in some kits I manage. Same surname and paper trail leading to a common ancestor from the late 1500's through different sons. Some don't even show up as Y37 matches due to having a GD greater than 4.

                Of course there are matches that could be NPEs with the same genetic distance. SNP testing through a NGS test like Big Y will remove any doubt.

                Two of the 3 markers you listed are labeled as fast mutating by FTDNA. If you look in a project's results page, the markers are labeled at the top. The ones in red have been determined by FTDNA to mutate more frequently.

                Based on what you have said, it looks good.

                Comment


                • #9
                  The_Contemplator

                  Thank you very much; I don't think I had realised how much variability there could be in the data! I draw the conclusion that these yDNA tests are almost like paternity tests... they can't prove that you are related but they can prove that you're not... except now the answer is 'you might be' :-)

                  Cheers
                  Karen

                  Comment


                  • #10
                    Generations Probability Cumulative
                    1 0.001317 0.001
                    2 0.007626 0.009
                    3 0.018626 0.028
                    4 0.031953 0.060
                    5 0.045167 0.105
                    6 0.056488 0.161
                    7 0.064923 0.226
                    8 0.070142 0.296
                    9 0.072285 0.369
                    10 0.071770 0.440
                    11 0.069144 0.509
                    12 0.064976 0.574
                    13 0.059798 0.634
                    14 0.054062 0.688
                    15 0.048133 0.736
                    16 0.042286 0.779
                    17 0.036716 0.815
                    18 0.031551 0.847
                    19 0.026863 0.874
                    20 0.022682 0.897

                    Using the 30 year generation convention, they're probably all around 6 gen. out from the common ancestor. So roughly 16% percent chance. Not encouraging, but not out of the question.

                    http://clandonalddnaproject.org/inde...rca-calculator

                    Comment


                    • #11
                      Originally posted by benowicz View Post
                      Using the 30 year generation convention, they're probably all around 6 gen. out from the common ancestor. So roughly 16% percent chance. Not encouraging, but not out of the question.

                      http://clandonalddnaproject.org/inde...rca-calculator
                      I have a match at 37-markers with a distance of 3 (34/37). Using the TiP calculator, this gentleman & I are predicted to share a MRCA within 12 generations with a probability of 90.31% (within 24 generations at 99.82%).

                      Using the MRCA Calculator from the Clan Donald USA site, set for 37-markers distance of 3, I get an estimate of 9 generations at .072. That's without Cumulative Probability because I didn't know what that meant.

                      We suspect a NPE in our paternal lineage and nailing down the exact date of our MRCA would help in proving or disproving our theory. Which is more reliable, the TiP Calculator or the MRCA Calculator?

                      Comment


                      • #12
                        Actually, that chart I copied and pasted fits the fact situation you present of 2 donors with GD of 3 at 37.

                        The Macdonald calculator uses very slightly slower mutation rates than the TIP calculator. I think the TIP calculator may be using more up-to-date information, but the differences aren't normally enough to get me bothered for such a rough estimate.

                        The percentages posted in my chart are cumulative. 36.9% at 9 generations. The 0.072 % figure that you quote seems as if it is the non-cumulative figure. Apples and oranges.

                        I only ever pay attention to the cumulative figure. You can look up and compare the difference between the technical definitions of cumulative and non-cumulative risk if you want, but a good, safe approach is to go with the cumulative percentage. I think most people only pay attention to the cumulative.

                        Either TIP or Macdonald are good choices. TIP's mutation rates may be more up-to-date, but Macdonald's are close enough, and probably considered more conservative. Which I guess is good.

                        Comment


                        • #13
                          >I don't think I had realised how much variability there could be in the data!

                          Yes, that's the way it often is. The issue is that over a short range of generations there can easily be zero, one, two or three changes. Over a longer range of generations all this variability smooths out, and we get the simple picture that all the introductory texts imply is normal. Over the short range, variability is normal -- not helpful, but normal. :-)

                          I agree that BigY testing of everyone involved would probably help resolve this - at a significant cost -- although there are no guarantees. Essentially you would be testing thousands of markers, instead of just a few dozen, and can even find new markers that are specific for your lineage. There is a good chance you'll be able to find among all these markers a set that lets you build a tree showing which markers arose in which sub-lineage. However, that still won't necessarily tell you in exactly which generation a change appeared, and therefore won't necessarily tell you who is a sibling and who is a first cousin, for example.

                          So a less expensive way to begin is to do BigY testing of one person, and find out if there are possible family-specific SNPs (private SNPs) in that lineage that give good quality (unambiguous) results in testing -- many private SNPs turn out to be poor quality, poor prospects for testing. If that first BigY produces promising lineage-specific SNP candidates, you can do another person's Big Y to narrow down the set of diagnostic SNPs to the most promising ones, or go straight to single SNP testing (US$17 ea. at YSEQ for custom tests) of the other relatives if there are just a few good candidates for lineage-specific SNPs. That's likely to be less expensive than BigY on everyone.

                          If anyone else in your terminal haplogroup but outside your lineage has already done BigY, then it will improve your odds of finding good lineage-specific SNPs because you can exclude the ones in common with that person. Check your haplogroup on the YFull tree for that.

                          Comment

                          Working...
                          X