When an autosomal match is not a match

[cjm]

Have you ever been sent an email inquiry about an autosomal match and after reviewing your tree and all the possibilities, you just come to the conclusion with the other person that either a. the match is just a random DNA segment match...we aren't actually related or b. we may be related but it's probably not in our paper trails so it's kind of a meaningless match? I've had a couple inquiries recently where I've just come to the concusion that this person couldn't possibly be related to me anywhere within a 5th cousin range or if they are, it's so far back, there's no way to make a meaningful connection.

[John McCoy]

The answer depends on the strength of the match -- the size of the matching segment in cM. If it's over 10, there's probably something there. If it's over 15, it's almost certainly real. However, the actual relationship could be quite a ways back. Autosomal DNA is transmitted in about 3 dozen chunks, as a result of recombination and the random segregation of chromosomes at each generation, so it is not the case that everything is completely scrambled. Rather, we expect some segments to be transmitted more or less intact through several generations simply by chance.

A couple rules of thumb bear on this problem: roughly 10 percent of actual 3rd cousins don't match each other by the usual default autosomal matching criteria, and about 50 percent of 4th cousins likewise don't match. So by the time we get out to your 4th cousins, quite a few of them won't have enough shared DNA to match you, even though some of them will match each other. But if many 4th cousins end up with too little shared DNA to match you, there must also be a lot of other 4th cousins who got MORE than the theoretical share of matching autosomal DNA. Some 4th cousins, and indeed some 5th, 6th, or even 7th cousins apparently can be identified among our matches. I've found some 4th cousins who share only about 9 cM with my father, but there are enough individuals involved, and a clear enough paper trail to convince us that we are on the right track to identify our mysterious McDougall ancestor.

In most cases, it will be very difficult or perhaps impossible to prove the identity the common ancestor when the connection is so far back, because few of us have complete pedigrees, and any of our pedigrees could potentially contain one or more errors. If, as happened with my McDougall family, there are multiple people who descend from different branches of both families, and enough of share the same segments, and if the pedigrees don't overlap in any other way, there may be enough evidence to lead to a solution..

When I get an inquiry about a single segment match, I look for an obvious connection in the family history, and I look for triangulation with other known relatives. Usually I don't find anything, and respond back that, while there is probably a distant relationship, we don't have enough evidence yet to reach a more definite conclusion. I'm afraid most single-segment matches are like that, but we keep trying, because every so often one of them can be identified.

[Germanica]

It depends how much DNA we share and whether we have any matches in common that I've identified. I had one person at AncestryDNA decline to give me an invite to her tree because she didn't think we were closely related enough to bother trying to figure it out (even though we share 24 cM in total across one segment). Years went by, and I finally identified the same MRCA of 3 matches I had in common with her. This helped me break down a brick wall in my tree and find the right records to confirm it. I message her to let her know that 4 her of matches are descended from this same MRCA so that's probably the branch we match on as well.

According to ISOGG, 15+ cM is supposed to have a 100% chance of being identical by descent. 11 cM has a 90% chance of being IBD. https://isogg.org/wiki/Identical_by_...sitive_matches

So I would definitely continue to pursue anything 15+ cM, and probably anything over 11 cM too (segment length, not necessarily total).

[cjm]

For me, one guy had 13.43 CM segment and one part of his family was from the same area as the suspected side, but we didn't connect on any names. I had to tell him that either it's a random match or we match probably somewhere in the 1600's...past our paper trails.

The other person, I was able to use DNA painter to find the exact DNA segment we were matching on - it was a 20.96 CM segment...an ancestor/wife born in the 1730's. But once again, none of his furthest ancestors matched anyone in my umbrella..like I mentioned before, I can pick up anywhere in a 5th cousin range with a furthest ancestor. I also didn't match that person's Y-numbers. So I said it was probably either one of the wives going back from the 1730's or it was a descendant of a brother or sister from my paternal line from off my paper trail. The person never responded back, probably because it was useless.

[Germanica]

For me, one guy had 13.43 CM segment and one part of his family was from the same area as the suspected side, but we didn't connect on any names. I had to tell him that either it's a random match or we match probably somewhere in the 1600's...past our paper trails.

The other person, I was able to use DNA painter to find the exact DNA segment we were matching on - it was a 20.96 CM segment...an ancestor/wife born in the 1730's. But once again, none of his furthest ancestors matched anyone in my umbrella..like I mentioned before, I can pick up anywhere in a 5th cousin range with a furthest ancestor. I also didn't match that person's Y-numbers. So I said it was probably either one of the wives going back from the 1730's or it was a descendant of a brother or sister from my paternal line from off my paper trail. The person never responded back, probably because it was useless.

Don't give up on the paper trail. My mom has a branch that goes all the way to the 1600s and sometimes even 1500s. Because these ancestors were a part of very small, early colonial communities, it means there is some endogamy. This means despite how far back these ancestors go, I have found dozens of DNA matches who descend from them, I think because the endogamy gives us a little more shared DNA. But not too long ago, this branch was a brick wall of mine, and I couldn't get passed my 3rd great grandfather. Had I never broken down that brick wall, all those dozens of matches I wouldn't have been able to find a connection to. There's probably dozens more who I haven't identified because their trees don't go back far enough. My most distant matches so far are 11th cousins - again, there's endogamy so we probably share more DNA than the average 11th cousins (we share around 13 cM), but my point is, you could have endogamous matches like this too, and not even know it if your tree doesn't yet go back far enough. I have a 9th cousin once removed who I share 24.5 cM with - again, there's endogamy, so that's not a normal amount to share, but I think people underestimate how likely endogamy is if you have early colonial ancestry so it's very possible some your matches are very distant cousins. I do think that makes it useless though, I have been working on my tree for 10 years and still break down brick walls. Every time I do, I go look for DNA matches to help confirm. And in some cases, it's the DNA matches that help me break down the brick wall!

Here's my list of identified matches, if you're curious (obviously, usernames were removed for privacy in my public version): https://docs.google.com/spreadsheets...c7sUOa/pubhtml

[Biblioteque]

Germanica - You have several "Shares less DNA with Dad/Mom, than me". Can you elaborate on that aspect of recombination? I think I have only seen that one time.

Re brick walls, what was the most amazing/interesting way you broke down a brick wall?

[Germanica]

Germanica - You have several "Shares less DNA with Dad/Mom, than me". Can you elaborate on that aspect of recombination? I think I have only seen that one time.

It's hard to say since AncestryDNA have no chromosome browser. I am speculating that Ancestry's Timber system which removes segments it deems irrelevant can be overzealous and inconsistent at times. All of us tested on chip v1 except my dad, but that's not the issue, as you can see I still have a lot of matches who share less dna with my mom than me. It could maybe be a matter of "no calls" but I thought I read that no calls are counted as a positive match by most companies so this shouldn't happen (plus I don't think no calls should account for more than one missing cM?).

But additionally, there is something not right going on with AncestryDNA's shared totals. Timber isn't supposed to apply to anyone closer than an estimated 3rd cousin I think - yet, my totals with my own parents are lower at AncestryDNA than they are at Gedmatch.

Re brick walls, what was the most amazing/interesting way you broke down a brick wall?

The most amazing was probably finding my grandfather's bio father. The above story was pretty interesting too (with the woman who didn't want to collaborate). I found 3 matches who all matched each other and looking through their trees, I found they were all descended from Jacob John Gossard. I knew I had a woman's maiden name in my tree that was Gossard but I didn't have her parents. I checked Jacob's will, and there was "my daughter, Catherine married to Peter Springer". There were already some trees that had her as his daughter, but with no sources attached I didn't know whether to trust them. I could have found this without DNA, of course, but DNA brought it to my attention and confirmed how I was related to the woman who didn't think we were closely related enough to pursue collaboration.

Other cool discoveries have actually been the opposite direction, finding and confirming other children of my ancestors. I once messaged a DNA match who had a John McBride in her tree and I suspected he was the John McBride Jr in my tree, son of my ancestors John McBride Sr and his wife Catherine Grosskopf (I had his baptism record). She wasn't really convinced, because there was so little recorded evidence of her John McBride to begin with, she only knew his name from some orphans court records of his children because he died while they were minors. She didn't know when he was born or died or anything else about him, so she was reluctant to make a connection between the two based on the name alone, especially considering John is a super common name and McBride isn't unheard of. So I scoured the records and found in Catherine's probate records, although she never mentioned her grandchildren by name in the will, the other probate records mentioned/are signed by all the names of her grandchildren by John McBride Jr, which were the same as the children of John McBride in my match's tree. They are not identified as Catherine's grandchildren (which is why I didn't know who they were), and John Jr was never mentioned because he was deceased, but all of the grandchildren's names listed there in the probates, plus the DNA connection finally convinced her. It was like she had one half of the puzzle and I had the other half and we never would have put them together if DNA hadn't brought it to our attention.

[Biblioteque]

Germanic - Thank you for sharing that information. Yes, Ancestry does have growing pains, and is sometimes a PAIN. This week I had a 6.9 cMs match with someone with present Ukraine/Russia ancestry. That was quite a surprise and I was intrigued enough to message him. We exchanged Gedmatch kit numbers, and he responded we do not share Dna at Gedmatch. I told him to drop the threshold, and he would find it at 3 cMs. He was not impressed and I did not hear from him again. LOL

But, I do have matches at Ancestry that do not show ANY shared Dna at Gedmatch. I do also have a number of 6-7 cMs tree matches at Ancestry.

[Germanica]

Germanic - Thank you for sharing that information. Yes, Ancestry does have growing pains, and is sometimes a PAIN. This week I had a 6.9 cMs match with someone with present Ukraine/Russia ancestry. That was quite a surprise and I was intrigued enough to message him. We exchanged Gedmatch kit numbers, and he responded we do not share Dna at Gedmatch. I told him to drop the threshold, and he would find it at 3 cMs. He was not impressed and I did not hear from him again. LOL

But, I do have matches at Ancestry that do not show ANY shared Dna at Gedmatch. I do also have a number of 6-7 cMs tree matches at Ancestry.

Yes, Ancestry includes matches below Gedmatch's default thresholds. Even Gedmatch's default of 7 cM has a 58% chance of not being identical by descent so I generally don't spend much time on a match below that (if a MRCA is obvious, then of course I make a note of it).