This is a Message Board Post that is gatewayed to this mailing list. Author: ksaxe Surnames: McConnell, McDonald Classification: queries Message Board URL: http://boards.rootsweb.com/surnames.mcconnell/2482/mb.ashx Message Board Post: Known Cousins Mismatch on 37 Markers John C McConnell and Robert A McConnell are both descendants of a man named Guian McConnell who was probably born between 1725 and 1740. John descends from Guian's son Alexander (b. 1787) and is just 5 generations removed from Guian. Both Robert and John have been tested on 37 markers, and with just 12 generations between them, they certainly expected to match on 37 markers. What happened? Well, FTDNA's match page and e-mail notifications did report them to each other as matches on 12 markers and 25 markers, but not on 37 markers. After all, isn't a 37 marker test supposed to give a more accurate picture of relatedness than a 25 marker test? It is, and to figure out what is going on here, it helps to look at the mismatches. Marker Robert John 449 30 31 464c 16 17 576 17 20 CDY B 40 41 The first two markers are in FTDNA's second panel of markers (13-25) and the remaining two are in the third panel (26-37), so Robert and John match 12/12, 23/25, and 33/37. 33/37 matches are often reported as matches, but one of the mismatches here is a 3 step mismatch, and when the total number of steps is calculated, the result is 6. The 6 step distance leads to their being classified as a mismatch by the software. What about knowledgable humans looking at the situation? Well, experienced people that I have talked to about this have shared some ideas with me, and these ideas generally lead to the conclusion that a mismatch like this is unusual, but not extremely rare. Sometimes a single mutation can cause a multiple step change in a marker. These mutations are supposed to be rarer than single step mutations, but a single multiple step mutation is much more likely than multiple single step mutations on the same marker. Among individuals who have a multi-step mismatch and agree closely on other markers, it's generally best to assume that a multi-step mutation has occurred and take the contribution to genetic distance from that marker as 1 instead of the actual number of steps. This is the approach taken in the Clan Donald project results tables, and when you click on John's and Robert's Clan Donald codes to compare their results, you find that they are each other's closest matches, with a genetic distance of 4 and a an estimated Time to Most Recent Common Ancestor (TMRCA) of 14 generations. This is still over twice as many generations as the average of 6 back to Guian, but it is in a range where you would expect to find results match. Another approach is to say, hey, 3 step mismatches are quite unusual. There may have been a mistake made in this case, and you should ask FTDNA to review their results and retest on that marker if there is any ambiguity. A review has been requested and FTDNA has agreed to look at the results, as they have done for others with unusual results. The results are pending, and in the meantime we treat the mismatch on 576 as the result of a single multistep mutation. Well, OK, you say, the 25 step mismatch does seem to indicate that John and Robert are related, but even treating the mismatch on 576 as a single step mutation still yields a TMRCA of 14 generations, when a value closer to 6 might be expected. What are the chances of that? Is is very much more likely than John and Robert not being members of Guian's family and matching 23/25? In other words, is it more likely that their paper trail is completely wrong and their match is due to a much more distant common ancestor than Guian, or that the paper trail is right and they just have a rather high mutation rate in their lines? I'm discounting the possibility that John and Robert descend from a common ancestor who was closely related to Guian, because that's a more complex situation and Guian was an immigrant, so it's unlikely that any such error in the paper trail would add more than a generation or two to their lines. In that case, the degree of mismatch in their results wou! ld still be somewhat greater than expected anyway. Project co-administrator Doug McDonald calculated a probability of two men with 3 1 step mutations and one multistep mutation on 37 markers being at least as closely related as John and Robert as 2%. That's a low percentage, but not extremely low. It's also possible that the 2% figure is a little low. He based his calculation on an average mutation rate of .0033 mutations per marker per generation, which is on the low end of current estimates. FTDNA uses higher mutation rates and calculated a 22.76% probability that John and Robert shared a common ancestor born within the 200 year period prior to their births, and a 51.76% probability that they shared a common ancestor born with 300 years of their births. John and Robert are of different ages. If you take their average birthyear and use that, you find that the ~ 23% figure applies pretty well to them. The true probability of men with such a match being as closely related as John and Robert are probably lies between 2% and 23%. The probability of 2 men who don't share a common ancestor within a genealogical time frame having such a DNA match is much smaller than 2%. In the case of Robert and John, they have rare haplotypes and very few matches. In fact, I don't think they have any 37 marker matches other than each other and using Y-Search to search for 25 marker matches yields only a few matches from many thousands of people. So an accidental match in this case is extremely unlikely. John and Robert already knew that they were related, and didn't need FTDNA to put them together, but if they hadn't known of their relationship, they would have had to be careful to recognize the evidence of a relationship provided by their test results. Next time I'll share another example of such a mismatch within the project and some tips for finding, recognizing, and understanding these possibly related "mismatches" in available databases. Important Note: The author of this message may not be subscribed to this list. If you would like to reply to them, please click on the Message Board URL link above and respond on the board.