No announcement yet.

Second opinion needed

  • Filter
  • Time
  • Show
Clear All
new posts

  • Second opinion needed

    I received the following e-mail asking for help in interpreting Family Finder results.

    I would like to ask you a question. My question is as follows:

    Is there scientific validity to concluding that specific large chromosome segments on which 25 male and female people in the same surname group match each other, are DEFINING chromosome segments for that surname? Note that the people in this surname group have a documented common history that goes back to the 1400's.

    Here is a description of the specific case which is of interest to me:


    · ~12 Halperns did the FamilyFinder DNA test.

    · 13 Alperoviches, where Alperoviches know that they are a branch of the Halpern family that broke off in about 1700, also did the FamilyFinder DNA test.

    · These two Halpern GROUPS match each other on the following chromosome segments:

    o Chromosome 1: Segments 115M-157M, many matches especially in 144M-149M, without a break, even for the large segment described.

    o Chromosome 6: Segments 10.8M-109M, with breaks.

    o Chromosome 18: Segments 11M-76M, with breaks.

    · Also, Halperns match each other in the same way on the above chromosome segments.


    1. Can the above specified chromosome segments be DEFINED as HALPERN chromosome segments?

    2. Can non-Halperns who match Halperns on any single part of these segments be DEFINED as Halperns???

    3. Can non-Halperns who match each other on any single part of these segments be DEFINED as Halperns???

    If these questions are not in the province of your interests and espertise, can you suggest to me someone with suitable credentials who can answer these questions ?

    And here is my reply:

    First of all, I don't think that I have the expertise to give you a definitive answer to your questions, although I can offer some ideas to consider. Although I'm a customer and project administrator at Family Tree DNA, I haven't taken the Family Finder test myself. I have tested at 23andMe, which has the Relative Finder feature, which is very similar to Family Finder. So, I'm very familiar with the basic principles of using autosomal DNA test results to find shared segments indicating a common ancestor.

    There are a couple of complicating factors in the information you presented below. The more significant measure of shared segments is not mb, which you've used below, but centiMorgans (cM). Family Finder should have given each of the 25 people you're asking about the total shared DNA in cM, plus the longest segment in cM. Can you give me some of those cM figures? That would tell more about the matches than citing the mb locations.

    The other factor is that I'm guessing, since you're writing from Israel, that these two family groups have significant, if not majority, Ashkenazi Jewish ancestry. Because of hundreds of years of intermarriage among a small population of Ashkenazi Jews, analyzing autosomal DNA test results is trickier. Basically, anyone with significant Ashkenazi ancestry is a distant cousin of virtually all others with significant Ashkenazi ancestry. So, this throws off the analysis, since shared segments are shared across a wider range of people than among non-Ashkenazi people. The long period of intermarriage has preserved segments for longer periods of time and distributed the same segments through wider circles.

    One thing you might try is using the "in common with" feature in Family Finder. This allows you to assign a relationship to those with whom you share segments and then you can see which of your matches also match each other, including specific segments. If you don't know the exact relationship, you can temporarily assign the relationship as "distant cousin" so that you can use the "in common with" feature.

    Given what I've written above, to answer your questions:

    1. The chromosome segments you've specified probably can be identified as Halpern segments, although the Ashkenazi factor I discussed clouds the answer.

    2. Again, because of the Ashkenazi factor, I don't think you can say with much certainty that non-Halperns who match those segments are actually Halperns. The segments may just be very common in the Ashkenazi population and not indicate a close relationship.

    3. Same answer as #2.

    If anyone can give a better insight into how to answer this person's questions or needs to correct anything I've written, please do so.

  • #2
    Unfortunately the answer is no to this person's three questions.

    If we assume the original common ancestors have only two children, then the descendants of the two children will match in the same locations. Eventually as each generation passes the descendants will no longer match due to DNA recombination.

    If on the other hand, the original ancestors had three children, or more, or had more than one spouse, the descendants will match sometimes in the same locations and sometimes in different locations. The reason has to do with the fact that the reference points of the comparisons change.

    Starting with two ancestors, when they only have two chldren then we are comparing all of the descendants of child A to child B; A to B in other words. If the original ancestors have three children then we are compareing A to B, B to C and C to A and in that situation, each comparison will be different. It gets even more complex with more than three children of the ancestors.

    Hopefully this somewhat clarifies the fact that many people may be related but not match at the same locations. It just depends on whether all the people we are comparing are descendants of the same two siblings or not.

    Of course this said, it is possible many people are all related and will match at the same location. The problem is that we can't say with 100% certainity that every time many people match, they are matching by the exact same ancestoral DNA segments. Every situation and every personal comparison may be different.

    The next issue, as you brought up, is the possbility of endogamous matching. When people are interrelated, the test begins to detect multiple ancestral matches in the same two people. When this happens it is possible that some segments are by one set of ancestors and other segments are by other ancestors. This would make it more complicated to say that certain segments are only from speicific ancestors.

    The problem is simply trying to say people will always be related if they match on a certain segment. Of course we see both paternal and maternal segments on the same map so even with that people may have an overlapping segment and not be related. For example my children match both grandmothers in some of the same places but neither grandmother is related to the other, its just that the grandchildren have paternal and maternal segments appearing on the same chrmosome map.

    The fact that people have matching DNA segments usually (but not always) means they inherited them from someone or more than one someone. The match is the clue, not the location of the match.

    The only valid method we can use to "MAP" DNA segments is to work backward from the present day to the past on an INDIVIDUAL basis using a paternal map and a maternal map. We start this by plotting two grandparents per map and then working backwords.

    Last edited by mkdexter; 17 September 2012, 03:55 PM.


    • #3
      I agree with Matt on base logic that any segment of DNA can originate with either parent and when dealing with the time frames in question may also be a sequence common within a population and would be difficult to say originated with a single individual or couple, especially if they may not have identified all ancestors back to the time frame in question or identified all potential NPEs.

      The base question attempting to tie segment of autosomal DNA to a specific surname (like Y DNA) appears flawed. Even if individuals can trace segment back to common ancestor or ancestor couple at a point in time, and rule out other possible sources, going back one more generation opens question of whether it came from their mother or father.

      Curious how many of the 25 persons tested actually share the segments in question and what other matches they are finding with primary segments in these locations in FF. Presumably, the majority would not carry the segments in question. If majority of them do, then you're probably dealing with something tied to an endogamous population with some sort of founders effect.


      • #4
        Originally posted by Geneadict View Post
        I agree with Matt on base logic that any segment of DNA can originate with either parent...
        Cousins of the same degree relation can share compounded segments originating from two full siblings. Cousins of different degree relation are the ones that share via one parent or the other.

        For example two cousins have a longest block match of 25.01cM. One matches me by longest block of 24.84cM and one by 17.56cM which is because I match this one a generation farther back; by one parent of full sibs. My 24.84cM match is really a compounded segment (father and mother) but the 17.56cM is just the mother's part of the same segment.

        The mother's side is a "Wright" by the way.... Oooooooh No

        Last edited by mkdexter; 20 September 2012, 01:49 AM.