Testing the cousins
One piece of advice you hear over and over in autosomal DNA testing: test as many relatives as you can afford to test.
And boy did The Legal Genealogist get handed the best evidence in the world as to why that’s good advice this past week.
Now before we go on, the reminder: an autosomal DNA test is the kind of test that works across genders and helps you find cousins in recent generations.1 Unlike YDNA, you don’t have to locate sons of sons of sons to test and only get results in the male line,2 and unlike mitochondrial DNA, you don’t have to locate daughters of daughters of daughters and only get results in the female line.3 With autosomal DNA, you can test the son of a daughter of a son against the daughter of a son of a daughter and get good results even across genders.4
But while you can test more people against each other to compare their autosomal DNA, you also have a shorter time frame for the results to be useful: unlike YDNA or mtDNA, which can be basically unchanged for hundreds, even thousands of years, autosomal DNA changes very rapidly from generation to generation through a process called recombination.5
In a nutshell, each parent has pairs of chromosomes but will only pass one of each to you. Before that happens, each pair the parent has gets mixed together and jumbled randomly so what you end up with may be more or less from any one ancestor. That’s why even brothers and sisters don’t have the same DNA: though each child gets 50% from each parent, each won’t get the same 50% as any other child.
Because of these changes, you get a rapid drop-off in the odds that you and a relative will actually share enough DNA in common to show up as matches. With a close relative, say through second cousin, the odds are pretty close to 100% that you share enough to match. Third cousins will match about 90% of the time. By the fourth cousin level, your odds are only 50-50; nine out of 10 fifth cousins on average will not share enough DNA to match; and by the sixth cousin level it’s really a crap shoot.6
Now 90% odds at the third cousin level are pretty good. So I was really delighted when, recently, we identified a third cousin in my Cottrell line.
He and I share a great great grandfather — my nemesis George Washington Cottrell.7 I descend through G.W.’s oldest son, Martin Gilbert Cottrell. This cousin, whose first name is David, descends from the youngest son, another George Washington Cottrell. His grandmother Luda Pearl Cottrell and my grandfather Clay Rex Cottrell were first cousins; his mother and mine second cousins; so he and I are third cousins.
But David had tested with AncestryDNA… and so have I… and we don’t show up as matches there. He matches a Cottrell cousin who’s also tested with AncestryDNA but not me.
I couldn’t believe it. I asked him if he’d be willing to test again at Family Tree DNA where the robust analysis tools would tell us so very much more than the “your tree matches his tree” (or not) results at Ancestry. He was willing, the test went in, the results came back this past week and…
He and I don’t match.
Now lest you sit there knowingly smiling to yourselves about some unreported non-paternity event, let me hasten to dissuade you from heading off down that blind alley. David matches both of my Cottrell uncles and my Cottrell aunt. He matches my Cottrell first cousin. He matches both of my Cottrell-side second cousins. He even matches my nephew, for cryin’ out loud.
He just doesn’t match me.
In other words, the key pieces of DNA that allow a positive match between this particular cousin and his relatives in my direct line happened to get passed down to just about everyone in my family… and to miss me.
To see this, take a look at the graphic with this blog post. You can click on it to see it bigger. This view of the chromosome browser shows in orange how my nephew and I match. You can see all those orange-colored chunks on just about every chromosome. These are chunks he would have gotten from his mother, my sister.
Now look at the chunks in blue. That’s where my nephew and my third cousin David share those very particular Cottrell segments that allow the testing company to be able to declare them a match. You can see that there is no orange in those areas.
David got those DNA segments. My nephew got those segments. My aunt, my uncles, my first cousin, my second cousins — they got those segments.
I didn’t.
In a situation where the odds are 90% in my favor… I landed in the 10%.
Maybe the DNA I got in those areas came from my grandmother and not my grandfather. Maybe it came from a different side of my grandfather’s family than the side that shows up here. Whatever the explanation, my DNA in those particular key locations just isn’t the same.
Now think about that for a minute.
If I was the only member of my family who’d been tested, and I was just starting out with autosomal DNA testing, can you see how I could have been misled by these results? It would have been so easy to think David and I weren’t related just because, in the random genetic crapshoot of autosomal DNA, the DNA dice just didn’t happen to roll the same way for the two of us.
I wouldn’t have known that it was just my DNA that differs, and in just the right (or wrong) place to make me the one who didn’t get the segments all my other relatives did.
So the moral of this story is: widen the net. If you’re going to do autosomal DNA testing, test all the relatives you can afford to test.
And don’t be discouraged when — not if, but when — you and a third cousin just don’t share enough DNA in common to be declared a match.
It happens to us all.
SOURCES
- See generally Judy G. Russell, “Autosomal DNA testing,” National Genealogical Society Magazine, October-December 2011, 38-43. ↩
- See ISOGG Wiki (http://www.isogg.org/wiki), “Y chromosome DNA test,” rev. 21 Jan 2013. ↩
- See ISOGG Wiki (http://www.isogg.org/wiki), “Mitochondrial DNA test,” rev. 29 Jan 2013. ↩
- See generally Russell, “Autosomal DNA testing.” ↩
- See ISOGG Wiki (http://www.isogg.org/wiki), “Recombination,” rev. 12 Jan 2013. ↩
- Russell, “Autosomal DNA testing,” at 39. ↩
- See e.g. Judy G. Russell, “Oh George… you stinker!,” The Legal Genealogist, posted 9 Jun 2012(https://www.legalgenealogist.com/blog : accessed 6 Apr 2013). ↩
Thank you for the informative piece. I am pretty new to the area of DNA testing and found this explanation easy to understand as well as interesting.
Thanks for the kind words, Vicki. I’m glad it’s useful to you.
Hello, cousin! Who knows if you’ll ever see this, but you and I are straight up 11th cousins according to the app on my phone. Of course, there are probably a million of us, aren’t there?
11th cousins! Well, who’da thunk it? What line of descent?
I have a 5th cousin on my Sowder line that did not match me when we tested on Ancestry.com. However, we both matched her Sowder second cousin. As we kept comparing we were both matching with different people that happened to go back to the same Sowder line, one we had thought was unrelated (and may be, but we know now to investigate the possibilities further). We also have a different set of matches that go back to a Riggs line we know we share. So the two of us having ended up with different segments of that Sowder line DNA has doubled our pool of matches.
And it’s every chance at doubling those matches that we’re trying for, Nan!
Here’s another way to convey the issue. Both my sister and I have tested on 23andMe, and we each have about the same number of people in our Relative Finder lists (DNA Relatives under the new 23andMe interface). But only about half of those people are common to both of our lists. So by testing her, I increased my number of relatives by about 50%, though I share no appreciable amount of DNA with the additional 50%. That’s not to say I can find genealogical connections to any of these people, but that’s a rant for another day.
Great example, Ruy! Thanks.
Judy, this is the most sensible and easy-to-understand explanation of how and why autosomal DNA testing “works” – and doesn’t. Thank you for sharing your experience with us and for explaining it to lucidly!
Kate
Thanks, Kate! Glad it’s helping make things clear.
Judy,
I cannot agree with you more just before I read your post I posted this on a thread at 23andme:
“I was looking at my son’s X (which he inherits from me) and comparing with my daughter’s two X’s. Well I can see that my daughter got the majority of her X from me, from my mother, and my son got nearly all of his X from that which I got from my father so if you were to compare these phased Xs my children hardly match at all! Now who would have guessed that?
And similarly I have two segments of African. My daughter got neither and my son the smaller of the two so in another generation all evidence of a fairly recent African ancestor (1700’s) will have disappeared.
Unfortunately for me my parents are deceased and no living siblings but even First and second cousins are helpful.
Kelly
The random distribution sometimes makes us all stop and say, “huh???”, doesn’t it, Kelly?
Judy, isn’t the moral of the story always “widen the net” (whether you’re working the paper trail or the DNA studies)? I seem to be saying that a lot to my genealogy students these days.
It’s sure the usual moral of the story, Lois… except for those times when you have too many people and need to narrow things down!
Sure do wish the prices had been better when I bought all 6 of my kits before. Now I have no excuse though. Still have my Dad and two granddads to test. If only ftDNA could get their act together so I could get back to work on those 6 kits…
Thanks for the insight. Funny about your “widen the net” phrase. I had said “cast a wider net” in my thesis today when I was describing why to take an autosomal test. Great minds think alike!
Sure wish any of my grandparents was still alive, Ginger! Grab those tests while you still can!
I have had the same success very recently by ‘widening the net’. A maternal uncle just tested and matches with known 4th cousins that my Mum and myself did not match with!
Glad to hear you had a bit of a breakthrough!
Just ran into this ‘Fun With atDNA’ myself. Possible 3rd cousin matched to my MIL’s kit but no match to my husband’s kit. Have to keep reminding myself that autosomal can rule someone in but doesn’t necessarily rule someone out. Still, at 3rd cousin level we should be able to work out the connection via genealogy.
Thanks for another great DNA post Judy!
Glad you’re finding this useful, Rorey, and yeah … it sure does remind us that DNA won’t solve all our issues, doesn’t it???
Great information Judy.
It absolutely tickled me that this new cousin matches you, Trish — and not me! EVERYBODY in my family got these genes but me!
Thank you for such a simple explanation of what I have been looking for from my ancestry matches. My one question is: If I was 100 percent Irish on Dads side and 100 percent on Moms side and The British Isles was found no where in my genealogy (both sides 200 years of Irish and Italian) what steps should I do next? Not sure I want to involve the Irish side in DNA testing. Could I get my raw data analyzed? Anyone to hire that you recommend? I think I owe my Mom an apology for doubting some things.
Lisa, first and foremost, remember that the percentages of British, etc., are guesstimates. The science just isn’t all that good now, so I wouldn’t be concerned that your results don’t show the kinds of percentages you expect. Even my Geno 2.0 results suggest that one of my key populations is Greek — and that’s just laughable. So ignore those percentages and concentrate on your matches, not the admixture guesstimates. You certainly can get your raw data analyzed and there are specialists out there who’ll do it. I don’t feel competent to recommend folks, but you can certainly get some good ideas and perhaps recommendations in places like the DNA-Newbie list on Yahoo or the Genealogy-DNA mail list at Rootsweb.
Help! I need your opinion!!!
The paper trail very strongly suggests that A, B, C, D, (siblings); E (their first cousin) and F (a second cousin of A, B, C, D and E) are third cousins of G and H (who are first cousins of each other). DNA tests return matches with G and H which are smaller than anticipated (based on ISOGG averages). BUT ABCDEF have matches with numerous other individuals with whom G and H also share matches. The surnames involved are not at all common in the parish in question. Should the fact that ABCDEF all share matches with numerous other individuals with whom G and H also share matches (albeit on different chromosomes or on different sections of the same chromosomes) be taken as supporting evidence of the original hypothesis (that ABCDE and F are third cousins of G and H? Given what is know with certainty – and given the rarity of the names – the paper evidence will not support any other conclusion.
Yes, it is part of your overall proofs on the issue. Remember that not all 3rd cousins will match; as many as 10% will simply not get enough DNA in common to show up as matches. So looking at the paper trail that says they’re 3rd cousins and the DNA evidence (all the 3rd cousins match a common set of others) you have a good case.
I just read this yesterday, and today I found out I have a 10% cousin, too! I was truly worried about one of my grandparents when a woman emailed me that she found a match in my online tree. She said her father tested for DNA – but he didn’t show as a match to me even though on paper we’re 3rd cousins. My mother’s results came back and they do match. For some reason, mom didn’t pass those genes on to me.