Announcement

Collapse
No announcement yet.

mtdna & FF I'm confused

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

  • #16
    Originally posted by Germanica View Post
    Yeah, I don't know why they stopped including the genetic distance chart - people are always asking what the genetic distance means and that orange chart doesn't help with that. They still have a genetic distance chart for Y-DNA, so why they removed the one for mtDNA is not something I understand and I'm just glad the original one is still available at ISOGG.
    I am guessing that evolutionary convergence ( https://isogg.org/wiki/Convergence ) could be the reason FTDNA does not include the mtDNA chart anymore.

    Mr. W

    P.S.
    Among my family members who tested at FTDNA, there are two whose mtDNA haplogroup is H. They have hundreds of matches, and if the table was valid there should be at least some from the same region in Europe. And there are none. I know, the sample size is small, but I am convinced that my guess goes in the right direction.
    Last edited by dna; 29th October 2017, 12:49 AM.

    Comment


    • #17
      Originally posted by dna View Post
      I am guessing that evolutionary convergence ( https://isogg.org/wiki/Convergence ) could be the reason FTDNA does not include the mtDNA chart anymore.
      But then why have genetic distance to begin with? Why use genetic distance, but fail to explain what it is with the chart? If the issue is that genetic distance can be misleading or incorrect, wouldn't it be better to just not use it at all, rather that use it but not tell people how to understand it?

      And wouldn't this apply to Y-DNA too? So why have the chart for Y genetic distance, but not mtDNA?

      In autosomal DNA, you can also have endogamy which can fool you into thinking a match is a closer cousin than they are but they doesn't stop companies from estimating the relationship degree.

      Comment


      • #18
        Originally posted by Germanica View Post
        But then why have genetic distance to begin with? Why use genetic distance, but fail to explain what it is with the chart? If the issue is that genetic distance can be misleading or incorrect, wouldn't it be better to just not use it at all, rather that use it but not tell people how to understand it?

        And wouldn't this apply to Y-DNA too? So why have the chart for Y genetic distance, but not mtDNA?

        In autosomal DNA, you can also have endogamy which can fool you into thinking a match is a closer cousin than they are but they doesn't stop companies from estimating the relationship degree.
        I think most of the confusion is due to the following two factors:
        * DNA analysis is non-trivial;
        * DNA science keeps evolving at a rapid pace.

        Genetic genealogy started being very simple, and genetic distance combined with probabilities of relationship is an easy to understand concept, both for mtDNA and Y DNA.

        Not only that originally nobody knew that evolutionary convergence plays such a big role when comparing mtDNA lineages, but there were just not enough people who tested their entire mtDNA to see the issue.

        The same with Y DNA, although availability of the Y-DNA111 test makes it less of a problem. Also a SNP test can help in deciding whether it makes sense to use STRs for a relationship calculation. That is probably why Genetic Distance stayed with Y DNA results despite evolutionary convergence clearly happening in the Y DNA realm ( https://isogg.org/wiki/Convergence ).

        You are right, endogamy increases uncertainty margins, but how many of us would like to see a statement that there is 90% chance that the relationship is 4th-5th cousin, and that the 90% estimate is valid in 97% of cases. (The above numbers are entirely bogus, I have no idea what the actual number ranges could be.)

        Even today, people just grab the first number they see, say a 91% probability of a relationship within 16 generations, and ignore the remaining percentage (9%).

        Or how many people just say what?!? upon reading reading
        Are there any exceptions or special cases in how our DNA is inherited that would affect my predicted relationship and range?
        https://www.familytreedna.com/learn/autosomal-ancestry/universal-dna-matching/exceptions-special-cases-inherited-affect-shared-dna/



        On the fundamental level, genetic distance (mtDNA & Y DNA STRs) and predicted relationship level (Family Finder) are probably just fine. It does not appear that there is an increasing percentage of people who have unusual results that require advanced analysis. It appears that there are much more testers, so the count of individuals with with unusual results (that require advanced analysis) keeps increasing.


        Mr. W

        P.S.
        The above is me guessing FTDNA thinking... With the Big Y test, FTDNA clearly wanted to move away from offering interpretation of the results, so I guess my guess has some merit :-)
        Last edited by dna; 29th October 2017, 06:18 PM.

        Comment


        • #19
          Originally posted by Germanica View Post
          Yeah, I don't know why they stopped including the genetic distance chart - people are always asking what the genetic distance means and that orange chart doesn't help with that. They still have a genetic distance chart for Y-DNA, so why they removed the one for mtDNA is not something I understand and I'm just glad the original one is still available at ISOGG.
          The genetic distance chart is not accurate - on average, the number of generations should be increased by about a factor of 10, but average genetic distances are not very meaningful for evaluating mtDNA matches. So it would be best to delete the chart and explain how to evaluate matches.

          mtDNA has a very slow mutation rate, on average about 1 mutation in 3000 years, but it is also highly variable. Some people can have several mutations in 3000 years and their matches will be much more recently related; other people have had no mutations in their maternal line in over 10,000 years and their matches can be very distantly related.

          There is also a problem that FTDNA counts some mutation hot spots and heteroplasmies in the genetic distance, and people who are very closely related on the maternal line are sometimes shown with a genetic distance of 1, 2 or more. So you really need to compare the detailed results to see what mutations are being counted in the genetic distance.

          I've been posting this information for years - I wish we could get FTDNA to provide a better explanation for mtDNA matches. mtDNA matches can be useful for genealogy in some cases but only if people understand how to interpret the results.

          Regarding genetic convergence - this is an issue for STR matches in y-DNA, not for mtDNA.
          Last edited by GST; 1st November 2017, 11:45 PM.

          Comment


          • #20
            Gail, thank you for outlining your position with which I totally agree! I hope FTDNA is listening.

            And, in the future, FTDNA should reflect more transparency/clarity in their graphics so people understand the difference in mtDna and FF testing; and will not mistakenly order mtDna testing when what they really need is FF. Your customers deserve this......

            I HAVE had full sequence mtDna testing, and I am an advocate. As Gail said, it can be very helpful for genealogy. E.g., it can rule in, or rule out, Native American on the maternal line.
            Last edited by Biblioteque; 2nd November 2017, 12:05 PM.

            Comment


            • #21
              Originally posted by GST View Post
              The genetic distance chart is not accurate - on average, the number of generations should be increased by about a factor of 10, but average genetic distances are not very meaningful for evaluating mtDNA matches. So it would be best to delete the chart and explain how to evaluate matches.
              I think the chart makes it pretty clear it's only an approximation and the number of generations are not absolute.

              Comment


              • #22
                Originally posted by Germanica View Post
                I think the chart makes it pretty clear it's only an approximation and the number of generations are not absolute.
                The number of generations in the chart are wrong by about a factor of 10. That is a very poor approximation. But the real problem is that, because of the large variability in the accumulation of new mutations, there is no average number that is meaningful.

                To use mtDNA results, one needs to look at the age of their subclade, count the number of extra mutations in their results, and estimate how recently they might be related to people who share those mutations. If your subclade is 6000 years old and you have no extra mutations, most of your matches are likely to be very distantly related, but if you have 5 extra mutations, your matches are likely to be much more recently related.

                We also need to ask FTDNA to improve the way heteroplasmies and common mutations are counted in the genetic distance. There are cases of people who are very closely related who are not shown as FMS matches because FTDNA counts common heteroplasmies as multiple steps in the genetic distance.

                Comment


                • #23
                  Originally posted by GST View Post
                  The number of generations in the chart are wrong by about a factor of 10.
                  How do you know that? Can you provide a resource that supports this?

                  Comment


                  • #24
                    Originally posted by Germanica View Post
                    How do you know that? Can you provide a resource that supports this?
                    For the mutation rate, the 2009 Soares et al. paper "Correcting for Purifying Selection: An Improved Human Mitochondrial Molecular Clock". Here is the key text: "The substitution rate for the entire molecule was 1.665x10-8 substitutions per nucleotide per year, or one mutation every 3624 years."

                    Behar et al. 2012 discuss the extreme variability in the rate at which mutations occur in different lineages, which they call "clock violations".

                    I'm a project administrator for several mtDNA haplogroup projects and have also looked extensively at the full sequence samples in GenBank, and there are many samples in subclades that are several thousand years old that have no additional mutations. One example is U5a1a1 which is estimated to be about 7000 years old and has been found in ancient remains dated to about 5100 years ago. There are over 150 samples of U5a1a1 that have no extra mutations, and they could share a common maternal ancestor with their exact matches more than 200 generations ago.

                    You can check the age of your subclade and then check to see how many additional mutations you have. You should exclude at a minimum 309.1C, 315.1C, 522.1A, 522.2C because they are extremely common and not useful for evaluating matches. A16182c, A16183c, 16193.1C(C) and C16519T are excluded from Phylotree and also have limited value. If your subclade was not included in the age estimates in the Behar et al. 2012 paper, you can calculate the average mutation rate for people in your subclade and use the spreadsheet from the Soares et al. paper to estimate the age of your subclade.

                    Comment


                    • #25
                      Here is another way to look at the results, for my subclade U5a2a1 which is estimated to be about 6000 years old, there are 185 full sequence samples in the U5 project and in GenBank. On average people in U5a2a1 have 2.26 extra mutations but the distribution is shown below, where 15 people have 0 extra mutations, 39 people have 1 extra mutation, etc. If you are one of the people with several extra mutations your matches are likely to be more recently related.

                      #mut, number of people
                      0, 15
                      1, 39
                      2, 64
                      3, 32
                      4, 26
                      5, 4
                      6, 3
                      7, 2
                      Last edited by GST; 5th November 2017, 01:36 AM.

                      Comment


                      • #26
                        This is the point where we disagree, a mutation could have occurred and then (later) the ancestral state could have returned. We think stability, but there was none.

                        mtDNA tree is full of back mutations.

                        I have read current papers, and yes clearly there is no more talk about parallel mutations (convergence). However, I am not convinced that mtDNA changes are so trivial that www.phylotree.org would not list any parallel mutations (at least I could not see any).

                        I am backing out of this discussion, as I need to learn more from up to date literature. Maybe one day someone will write a complete explanation of mtDNA for genetic genealogists.


                        Mr. W

                        Comment


                        • #27
                          Originally posted by dna View Post
                          This is the point where we disagree, a mutation could have occurred and then (later) the ancestral state could have returned. We think stability, but there was none.

                          mtDNA tree is full of back mutations....

                          Mr. W
                          Mr. W, I think you are 100 percent correct. I have certainly seen examples that appear to belie the notion of mutations being thousands of years back. Within a few months I hope my research will be able isolate some examples of mutations taking place within hundreds of years instead of the thousands of years generally thought.

                          Jack Wyatt

                          Comment


                          • #28
                            Originally posted by dna View Post
                            This is the point where we disagree, a mutation could have occurred and then (later) the ancestral state could have returned. We think stability, but there was none. mtDNA tree is full of back mutations.
                            Back mutations do occur in mtDNA but they are relatively infrequent compared to the total set of mutations that define a haplogroup, and when you test the full sequence different subclades will not converge in the same way that yDNA STR can converge. Convergence is really only a problem when using a small number of markers to compare results, especially in the yDNA 12 and 25 marker STR tests, but less of a problem when you test more markers. Convergence can be a problem with the HVR mtDNA test, for example, multiple haplogroups can be CRS in the HVR region, but FTDNA also tests additional coding region SNPs as part of the HVR test.

                            There are certain mtDNA markers that do back mutate very frequently and they should not be included when evaluating matches.

                            There are also cases in which the precise structure of the mtDNA phylotree can be uncertain because of multiple reversions at markers that mutate frequently, but this is a reason to discount these markers when building the tree. I don't think it is useful to use markers 152, 195 etc to define "paragroups" as is currently done in Phylotree. So back mutations can be a problem, but testing the full mtDNA sequence or testing the full Y sequence basically solves the problem.
                            Last edited by GST; 5th November 2017, 11:45 AM.

                            Comment


                            • #29
                              So confused

                              The more I read about mtDNA, the more confused I become. I thought the mtDNA would help me better understand my mother's maternal line, but I was told by an administrator that the mtDNA traces the family of my mother's maiden name. That just doesn't make sense, as my maternal grandmother has no dna from my mother's maiden name. I understand that you learn your haplogroup, but how does the test rule out Native American ancestry? Thanks for any help!

                              Comment


                              • #30
                                Native Americans have their own Haplogroups which will rule you in or out.

                                MtDna traces your mother's LINE, not her name, back in time.

                                From Roberta Estes, a list of NA Haplogroups. Hope this helps.


                                https://dna-explained.com/2013/09/18...l-haplogroups/
                                Last edited by Biblioteque; 10th November 2017, 10:11 AM.

                                Comment

                                Working...
                                X