7

With DNA information from FamilyTreeDNA, I have determined trangulation groups. I have chromosome segments of significant size (> 10 cM) over which A (myself) matches B (my known cousin) and A (myself) matches C (some third parties) and I've found out from my cousin that B matches all the C's as well over the entire segment.

This 3-way triangulation is considered extremely reliable and all the third party matches will almost assuredly be related to me and my cousin's MRCA (Most Recent Common Ancestors) and/or some of their ancestors.

I've assigned these matched chromosome segments in my DNA to have originated from my and my cousin's common ancestors.

So now that I have this information, my question is what happens with the people I have found that match me and some of the third parties on one of those segments, but don't match my cousin? Or the people who match my cousin on one of those segments, but don't match me?

Is there an explanation for this?


To make this clearer, here is a concrete example.

enter image description here

The yellow line with the green X's near the top is my uncle's match on Chromosome 1 with my known third cousin, we'll call Joel. The green X's on that line is the segment where my uncle matches Joel.

Note the green X's of the other 29 lines. The green X area are the exact segments where those people match my uncle AND where those people match Joel. (I got Joel's segment match list from him so that I can do this).

(The pink and blue areas are where that person only matches my uncle, or only matches Joel, but that's really not relevant for this question).

But since we have a segment between 170876383 and 183564808 on chromosome 1 for each of the 29 people who match both my uncle and Joel somewhere on that range, then that is three way triangulation (A matches B, B matches C, and C matches A) and that means that all 29 people are pretty well guaranteed to have a common ancestor who is either my uncle and Joel's common ancestor, or an ancestor further back.

I thought that once this Triangulation segment is arrived at, then anyone else who matches either my uncle or Joel anywhere on the green area must also be in this group with a common ancestor.

But then I found these people, each of whom were only in one of the match files:

enter image description here

The people in column B match either my uncle (Harry) or Joel in Column A. But if they are matching to Harry, they are not in Joels match list. And if they are matching to Joel, they are not in Harry's match list. A large part of those matches is over the 170876383 to 183564808 range where Joel and Harry match with green X's for their common ancestor.

Each of those people are in only one match list, either Harry's or Joel's. Some of the segment lengths are quite significant, and it's unlikely the 13.18 and 12.81 cM segments are not IBD.

So I am confused as to how this can be and I'm looking for possible explanations.

  • I think this may need yet more detail for clarification - are you saying that for a specific DNA segment on a specific chromosome, with a given start and end point, that A matches B, A matches C and B matches C, but that D (your other third parties) match A and C but not B? For exactly the same segment? – cleaverkin Jun 14 '16 at 17:51
  • Yes. I triangulated with all the people (C) who matched me and my cousin to get the segment identifying our common ancestor. Then to my surprise, I found other people in my match list (D) that matched me on that same segment but did not match my cousin on the segment. My cousin had similar people on his match list that matched the segment but did not match me on that segment. – lkessler Jun 15 '16 at 5:50
  • OK, that's (barely) within the realm of possibility UNLESS the people in group D also match people in group C. Do they? And, sorry to repeat, but we're talking about segments with the exact same start and end points, no variation at all? – cleaverkin Jun 19 '16 at 5:48
  • @cleaverkin - I've added an example of what I encountered. Maybe that will trigger an idea. – lkessler Jun 21 '16 at 3:53
3

I'm by no means an expert on genealogical DNA (I probably know less about it than you do). I do seem to spend a lot of time looking at aspects of it that make no sense, though.

That said, I'm going to go Holmesian here re: the impossible and the highly improbable. My guess is that you're looking at half-IBD matches here - your uncle and Joel, and the 3rd parties that match both, are all matching on one half of the chromosome 1 pair, and the others are matching either Joel or your uncle on the other half.

I know it seems unlikely that there would be matches within the same segment range on two halves of the same pair, but the alternative (that DNA segment matches aren't transitive) seems impossible.

I've seen similar phenomena in my own matches, where I've mapped a particular segment to a particular ancestor via triangulation. Then, I've seen what appears to be a match of an overlapping section of the same segment to someone with whom I provably have no "recent" (DNA timeframe) common ancestors.

Of course, if you're dealing with more than one matching segment here, then forget everything I just suggested.

| improve this answer | |
  • You've pinned down exactly the reason why I'm perplexed. I've got enough of these that there must be a logical explanation for it. – lkessler Jun 22 '16 at 6:20
  • I don' know if this helps in your case but I have a case of pedigree collapse several generations back (a paternal 4X g-grandfather and a maternal 4X g-grandfathers married each others' sisters). I'm starting to see indications that some of my 3rd cousins who also descend from one of these couples match more segments with me than I would normally expect from 3rd cousins. I suspect the pedigree collapse, but don't have enough data points yet. I haven't seen a case of the same segment on both chromosomes in a pair (yet), but it could happen. Again, unlikely, but.... – cleaverkin Jun 22 '16 at 17:11
  • The more I'm thinking about it, the more I believe your answer ( not your comment) is right. And it is not necessarily a rare thing either. There is a low but significant chance that both halves could have been passed down to Joel or my uncle. 1/16 of the DNA goes to a g-g-grandchild. The same amount comes from the other g-g-grandparent. There are 150 or so segments after that many generations, so there's a good chance several of those can be made up of parts of both halves of a g-g-grandparent pair. Almost convinced. A bit more thinking required. – lkessler Jun 24 '16 at 0:26
  • I continued research into this and discovered some likely possibilities I describe in a blog post: Triangulations and Missing a-b Segments, and I believe you are correct so I'm accepting your answer. – lkessler Sep 28 '16 at 4:27
2

Found another reason. These people only have enough total cM matching with Harry and Joel to be considered a match to just one of them.

Yes, they do triangulate on this segment with both Harry and Joel, but this may be the only segment, or one of just a few, that matches to one of them. And only people considered to be matches have their segments included in the segment match file.

| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.