## Triangulation conundrum

5

I tested both my mother "M" and her brother "D" at FTDNA. I was looking through D's list of segments today for triangulated segments, using the matrix utility to help, and I found the following

D matches Person A on chromosome 1 from positions 55980120 to 62971860

D matches Person B on chromosome 1 from positions 55980120 to 69360781

D matches his sister M (my mother) from positions 54879395 to 81490003

According to the matrix utility, M is also a match to A and B but A and B do not match each other. At first, after some thought I decided that D and M must be full matches on the relevant segment (i.e., they match on both of their chromosomes), and A matched one chromosome while B matched the other. But using David Pike's utilities and double checking with the graphic at GEDmatch, it is clear they are only a half-match.

Then I thought maybe the match between A and B existed but was too weak for FTDNA to consider them a match. But their overlap is over 10cM and is over the threshold that FTDNA uses (which I found by Googling).

So I am at a loss as to how this can happen. Any ideas?

Doesn't the answer https://genealogy.stackexchange.com/a/14658/1006 explain this case as well?

– Jan Murphy – 2018-09-02T07:46:21.643

1I don't believe so. That diagram showed a triangulated segment. Here, there is no triangulation, so it's a different scenario. I know the "two legs of a triangle" match can happen with just A, B, and my uncle. But when you throw in that they all match my mother on that same segment of chromosome 1, even counting for there being two chromosomes, I don't see how it can happen. – Barry – 2018-09-02T12:19:32.233

2

You say D and M are not fully matching on that segment and are only half matching. In that case, A and B cannot be matching on opposite chromosomes and they are either both on D and M's paternal side, or both on D and M's maternal side.

So I see 2 possible reasons why A and B might not match each other.

1. D and M are a generation or two ahead of A and B. e.g. If A and B are 1st cousins twice removed to D and M, then there would be a two generations difference. In this case A and B would be 3rd cousins with each other and they just might not share enough DNA to show up as matches to each other.
2. Anyone who is a 2nd cousin once removed or further has a small probability that they don't share enough DNA to show up as matches to each other. It just could be that D and A, D and B, M and A, M and B were passed down enough shared segments to appear as a match, but D and M randomly happened to not get enough.

You say "A and B cannot be matching on opposite chromosomes", so it sounds like you agree that the (>10cM, in this case) stretch of chromosome 1 where both D and M match A and that same stretch where D and M both batch B must actually be on the same chromosome in the DNA D and M (i.e., we're using the exact same strand of DNA in matching D to A as we are in matching D to B). So A has a copy of that strand, as does B. How, then, could A and B not show up as matches? They would have the exact same >10cM stretch of DNA, so according to FTDNA's posted threshold should appear as a match. – Barry – 2018-09-02T16:32:27.747

I thought for a moment that if A was related to D & M's father's father. and B was related to D & M's father's mother, then A and B wouldn't be related. But that's impossible with the triangulated segment because it only comes down one ancestral path. – lkessler – 2018-09-02T16:32:55.973

The > 10cM stretch you see on can include random bits of matching at either end. They can total enough to make the matches of some people over 10 cM when the the actual common segment is under 7.69 cM, which is the amount needed by FTDNA to be considered a match. And it could be that A and B's match doesn't have the extra randomness on the end to put it over the 7.69 cM. – lkessler – 2018-09-02T16:43:37.750

Man, I wish I A and B both had kits on GEDmatch so I could check what you say here. It seems highly unlikely to me that I would keep finding short matching segments positioned exactly where they would need to be to foul up my thinking, but I cannot see any other possibility. Perhaps someday I will find this happening with a pair of people who are also on GEDmatch so I can believe it by seeing. – Barry – 2018-09-02T16:51:08.720

If you are in contact with either A or B, ask one (or both) of them to send you their Chromosome Browser Results file. (Use the "Download all matches to Excel" link that is at the top right of FTDNA's Chromosome Browser page.) That will contain all their segment matches with everyone they match to. It will tell you if they match to each other and who else on that segment. You can then compare that to other people you also match on that segment and see who you triangulate with. – lkessler – 2018-09-02T16:59:47.020

I just found three people who all share segments with the same starting point with D. and over 10cM in each case, but according to the matrix tool, none of these three people is a match to any of the others. That seems impossible to me. I'm beginning to think that the 4% no-calls in my data are the culprit. Unfortunately, FTDNA says the kits are within their quality control specifications. – Barry – 2018-09-04T02:00:21.320

I agree with Louis. The matrix tool is a waste of time as you can’t do triangulation without asking one of them for their file.

So do yourself a favour and ask them. If they don’t reply then that’s it.

Unfortunately FTDNA doesn’t provide us with the right tools. – a4xrbj1 – 2018-09-04T12:59:05.897

1

because A and B match on the opposite strands of chromosome 1. One is on the paternal side, the other one on the maternal side (those are your maternal Grandparents).

That's why they don't match each other which is required for a triangulation.

BTW, please don't use the word "half-match". People might misunderstand that they are only half-siblings when you wanted to state that they only have an HIR and not an FIR at that locus.

Thanks for the HIR and FIR tip. Since M and D are HIR on that strand, they match, let's say, on the paternal chromosome only. Say A was the match to D on the paternal chromosome and B was the match on the maternal one. How, then, can FTDNA report M and B as a match on that same part of chromosome 1? It shouldn't be on M's paternal side, since that's where M and D match and B isn't matching D on the paternal side. But it shouldn't be on M's maternal side either, because M didn't get the same maternal side as D and B is matching D on the maternal side. – Barry – 2018-09-02T16:47:19.320