Announcement

Collapse
No announcement yet.

67 Marker Results

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

  • #16
    Originally posted by JC399
    I disagree. The extreme example of that is where you only compare the last 30 in someone you are related to and they all match even though you may be 18 generations apart. In that case your most probable common ancestor is 0 generations with a wide distribution curve. Not much useful information there.
    Well, here's the bottom line. Genetic genealogy is all about probabilities and random mutations, with some order thrown in with paternal lines and haplogroups and their subclades. There is limited predictability, but what there is, is useful.

    No one, probably including FTDNA, had any idea what the mutation rates would be on these 30 new markers. The only way to determine those rates is to get a good sample of people testing, especially those who already knew about matches with close relatives in the first 37. Now we have found that only a few markers, except in cases of rare individual mutations or certain subclades, tell us much. That does finetune our knowledge, which is helpful.

    So perhaps you should have waited until all this became clear. Or you could look at it from the viewpoint of what was learned with your help.

    Mike
    Last edited by MMaddi; 8 January 2007, 09:45 PM.

    Comment


    • #17
      Originally posted by MMaddi
      So perhaps you should have waited until all this became clear. Or you could look at it from the viewpoint of what was learned with your help.

      Mike
      I'm not particularly worried about it. I am mostly trying to understand the results. When I first saw them I thought it was a big mistake. Now that I know the mutation rates are very low I understand them more. Furthermore, I don't trust the FTDNATiP results for the 67 markers.

      Comment


      • #18
        Originally posted by JC399
        The extreme example of that is where you only compare the last 30 in someone you are related to and they all match even though you may be 18 generations apart.
        Yes, that would be stupid. So don't do it. The upper 30 markers are not designed for stand-alone use. They are meant to be used in conjunction with the lower 37. We have found, in practice, that the lower 37 can mutate so rapidly as to lead to 'coincidental convergence.' The upper 30 weed that out.

        Comment


        • #19
          Originally posted by lgmayka
          Yes, that would be stupid. So don't do it. The upper 30 markers are not designed for stand-alone use. They are meant to be used in conjunction with the lower 37. We have found, in practice, that the lower 37 can mutate so rapidly as to lead to 'coincidental convergence.' The upper 30 weed that out.
          The point is those markers bias the result to make one appear to be more related than they are for the simple reason they don't change in a genealogical time frame. Do the math yourself as I have and you will see. But then intuition alone should tell you it's not a good idea to make time predictions from lots of markers that don't change, whether they are in conjunction with a few that do or not. Most of us are not looking for relatives going back 3000 years or more, we are looking back in the hundreds of years time frame or less. For that you need as many fast changing markers as you can find. That's where the information lies. I can see how such a thing might weed some people out, but there is a price to be paid for adding so many non-changing markers.

          Comment


          • #20
            Originally posted by JC399
            The point is those markers bias the result to make one appear to be more related than they are for the simple reason they don't change in a genealogical time frame.
            First, many of the markers apparently have very low mutation rates but some seem to have very high rates. The overall average rate for all 30 "new" markers is probably not a whole lot lower than the average rate for the first 25.

            Second, there is no "bias" introduced by testing markers of any mutation rate. As long as the TMRCA calculations use an accurate mutation rate estimate, there is no bias.

            Just as a mathematical curiousity, the only inherent advantage of using fast markers is that you can test fewer of them. From the point of view of getting an accurate TMRCA estimate, testing 20 markers with an average mutation rate of 0.005 is no better than testing 200 markers with an average rate of .0005. And, as I said, the average rate of the 30 extra markers is not nearly that low.

            The problem with using only fast markers is that you end up with convergence, as Lawrence has pointed out, since the markers tend to have quite narrow allele ranges. Many of the markers in the 13-25 and 26-37 panels have very high mutation rates, and the 38-67 panel balances that out.

            Generally, from a purely geneological point of view, there is no point in ordering the 67 marker test unless you already have close matches with 37 markers. But the extended test is, in fact, useful to many surname projects in a genealogical context. The fact that it is also useful to some people in a genographic context is a welcome bonus.
            Last edited by vineviz; 9 January 2007, 09:48 AM.

            Comment


            • #21
              Originally posted by vineviz
              First, many of the markers apparently have very low mutation rates but some seem to have very high rates. The overall average rate for all 30 "new" markers is probably not a whole lot lower than the average rate for the first 25.
              Wait a minute. Just a few posts back I was told by an FTDNA employee they were all quite low. Now you are telling me some of them are quite high? Which is it? What are the mutation rates?

              Originally posted by vineviz
              Second, there is no "bias" introduced by testing markers of any mutation rate. As long as the TMRCA calculations use an accurate mutation rate estimate, there is no bias.
              As far an accurate mutation rate is concerned, there is no way of verifying this because that information is a state secret. In my case people who were not all that closely related (about 20 generations for 50%) at 37 markers have been shifted about 4-5 generations closer at 67.

              Originally posted by vineviz
              Just as a mathematical curiousity, the only inherent advantage of using fast markers is that you can test fewer of them. From the point of view of getting an accurate TMRCA estimate, testing 20 markers with an average mutation rate of 0.005 is no better than testing 200 markers with an average rate of .0005. And, as I said, the average rate of the 30 extra markers is not nearly that low.
              I don't know what they are and I have seen nothing published to tell me what they are, nor have I seen anything published to tell me exactly how they are being used in FTDNATiP. I have read some of Dr Walsh's old papers, but whether that is the way FTDNATiP works is a bit of a mystery. If you say some of them are high, I am at a loss to explain why I am not seeing that in my matches. Small sample size?

              Originally posted by vineviz
              Generally, from a purely geneological point of view, there is no point in ordering the 67 marker test unless you already have close matches with 37 markers. But the extended test is, in fact, useful to many surname projects in a genealogical context. The fact that it is also useful to some people in a genographic context is a welcome bonus.
              I already have close matches at 37. In my case, the extra 30 markers didn't exclude anyone. They only made more distant matches closer.
              Last edited by JC399; 9 January 2007, 12:50 PM.

              Comment


              • #22
                Originally posted by JC399
                Wait a minute. Just a few posts back I was told by an FTDNA employee they were all quite low. Now you are telling me some of them are quite high? Which is it? What are the mutation rates?
                DYS413, DYS481, DYS534, DYS557, and DYS572 all have mutation rates that most people would call "high". Not all of them have published father-son rates, but they show variability in surname and geographic projects on par with markers that FTDNA publicly asserts to be "fast".

                I think there are probably six or seven that have very low mutation rates (i.e. unlikely to show any mutations in a couple dozen generations), and the rest are probably slow to medium.

                At their recent conference, I think FTDNA said that the overall mutation rate they were using for the panel was in the 0.002 to 0.0028 range (I don't have my notes with me), which is not much different from the rate for the basic 12 marker test.

                Comment


                • #23
                  I still maintain. If your DNA does not match, you are not related.
                  We went back thru 9 fathers in our family and got a perfect match.
                  Mutations is a word used in bad research. Find a fifth cousin, have him tested and then we can have a discussion.
                  d

                  Comment


                  • #24
                    Originally posted by darroll
                    Mutations is a word used in bad research.
                    "Mutations" is also a word used in good research.

                    Comment


                    • #25
                      Originally posted by darroll
                      I still maintain. If your DNA does not match, you are not related.
                      We went back thru 9 fathers in our family and got a perfect match.
                      Mutations is a word used in bad research. Find a fifth cousin, have him tested and then we can have a discussion.
                      d
                      I have distant cousins (with proven genealogy) going back 6 generations with 1 mutation in 25 and another with 2 mutations in 34 (from Sorenson).

                      Comment


                      • #26
                        Originally posted by vineviz
                        DYS413, DYS481, DYS534, DYS557, and DYS572 all have mutation rates that most people would call "high". Not all of them have published father-son rates, but they show variability in surname and geographic projects on par with markers that FTDNA publicly asserts to be "fast".

                        I think there are probably six or seven that have very low mutation rates (i.e. unlikely to show any mutations in a couple dozen generations), and the rest are probably slow to medium.

                        At their recent conference, I think FTDNA said that the overall mutation rate they were using for the panel was in the 0.002 to 0.0028 range (I don't have my notes with me), which is not much different from the rate for the basic 12 marker test.
                        Thanks for taking the time to explain this. Now that we have established the mutation rates are comparable to the lower markers, I'm still wondering why all my distant matches show no mutations in the upper 30 markers. On ysearch I found my closest matches at 37 who also took the 67 marker test. Attached are the results from ysearch. We all have 6 mutations or less at 37, yet all share the same identical upper 30 markers.
                        Attached Files

                        Comment


                        • #27
                          Originally posted by JC399
                          Thanks for taking the time to explain this. Now that we have established the mutation rates are comparable to the lower markers, I'm still wondering why all my distant matches show no mutations in the upper 30 markers.
                          Most of your genetic distance is at CDY, DYS576, and DYS449 which are the three fastest markers in the 37 marker panel. The mutation rate of CDY is eight times the panel average, and is possibly the fastest mutating STR we have.

                          Comment


                          • #28
                            Originally posted by JC399
                            Now that we have established the mutation rates are comparable to the lower markers, I'm still wondering why all my distant matches show no mutations in the upper 30 markers.
                            Be grateful. My cousin was only 5 steps away from a known lineage at 37 markers; but at 67 markers, he is now 13 steps away.

                            Comment


                            • #29
                              WHY?
                              One source said that a single mutation takes hundreds/thousands of years.
                              Then:
                              Maybe someone can explain why mutations do not show up in all family lines?.
                              FTDNA shows a Y-DNA 35 out of 37 match, a family member.
                              It is starting to look like mutations do not take hundreds/thousand of years that I have read everywhere..
                              Would some brain please clarify this and I can quit hollering “Bad Paper Trail”.
                              One close match to an unknown showed 11/12 on the 12 point.
                              This same person had his 37 point test done and he matched 35/37.
                              I do not consider him a family member even though he says he is related.
                              Help Please.
                              d

                              Comment


                              • #30
                                Originally posted by darroll
                                WHY?
                                One source said that a single mutation takes hundreds/thousands of years.
                                On average, the most commonly used Y-STR markers have a mutation rate of something like .003. This means that, on average, each marker will change its value approximately once every 333 years.

                                But you don't test just one marker, you test several (12, 25, 37, or 67 at FTDNA). So, if you test 37 markers and each of them mutates on average once every 333 years then you would expect one of the 37 to mutate every 9 generations.

                                So, for a 37 marker panel it would be more accurate to say that a single mutation takes about 250 years.

                                And since we are usually comparing our results to a cousin, you must count the generations in both lines together. If two people have a common ancestor just 125 years ago, they probably do not have idential Y-STR results for all 37 markers.

                                However, the mutations do no have a fixed schedule. They occur at random, so sometimes they happen quickly and sometimes they happen slowly.

                                The classic example is a coin toss. A fair coin will land on heads about 50% of the time, but it does not necessarily alternate heads/tails/heads/tails/heads/tails etc. It usually will land in short streaks (HHTHTTHHTTHT etc.) but sometimes in long streaks (HTHHTTTTTHTH). The average is known, but each individual outcome is unpredictable.

                                Comment

                                Working...
                                X