Maximizing autosomal matches
There is a reason why The Legal Genealogist occasionally sounds like a broken record when it comes to autosomal DNA testing.
“Test as many people as you can,” the song goes, before the needle skips and it repeats: “Test as many people as you can.”
Autosomal DNA testing, remember, is the kind of testing you do when you take the Family Finder test at Family Tree DNA, the AncestryDNA test from Ancestry.com or the test from 23andMe.
It’s the test that looks at the DNA in what are called the autosomes, the 22 pairs of chromosomes we all have containing DNA randomly jumbled and passed down equally from our mothers and fathers that helps us locate cousins1 because it contains segments from many different ancestors whose DNA, by chance, managed to survive that jumbling process (called recombination).2
And because of that random jumbling process, even brothers and sisters do not have exactly the same autosomal DNA. One might get more DNA from, say, their paternal grandfather than the others; another might get more DNA from their maternal grandmother. So the only way to find as many matches as possible — and to find the very best matches — is … yep, you’ve got it … “Test as many people as you can.”
Let me show you what that really means, using my own family as an example. Because very little shows why this “test as many people as you can” mantra holds true better than a review of the autosomal results of four members of my family who’ve taken the autosomal DNA test.
All four are full-blood siblings, brothers and sisters of my mother, in age order my uncle David, aunt Carol, uncle Mike and aunt Patricia, called Trisha. And the matrix chart you see here shows you the amount of DNA, measured in units called centiMorgans or cM,3 that each shares with the others.
Now the first thing you notice about the chart is the variation between the sibling pairs.
• Mike and Trisha — the two youngest of my grandparents’ 12 children — share the most DNA in common, with David and Carol, the two oldest who tested, coming in a close second.
• David and Trisha, from opposite ends of the age spectrum, share the least in common.
• David and Mike, the brothers, share more DNA in common than Carol and Trisha, the sisters.
And remember: it just happened that way, by random chance. Testing four siblings in your family won’t necessarily produce the same pattern of results.
So what does this tell us? If we’d just tested David and Trisha, we’d have gotten fewer matches in common overall than testing Mike and Trisha. And there’s no way we could have known that before we tested all four.
In other words — repeat after me — “Test as many people as you can.”
And that lesson is really drilled home when you look at the match lists for these four siblings. In total, as of this past week, they have 1,681 matches in the database at Family Tree DNA.
And not even 10% match all four siblings.
Only 159 of the people who match any of my aunts and uncles match all of them. These obviously are the primary cousin targets for research: with autosomal DNA, what we’re looking for more than anything else is patterns and these match-everybody matches may help point the way. And even when you subtract the known relatives who’ve already tested — two nieces (one the daughter of one of the four siblings), one great nephew, one first cousin, three first cousins once removed, two second cousins, two second cousins once removed, and a third cousin, among them — we’ve got some serious candidates to look at.
Not all of them are high priority candidates, of course. The smaller the amount of overall DNA that a match has in common with the siblings, and the smaller the size of the largest shared segment, the farther back in time the common ancestor is likely to be — reducing the odds of ever being able to identify that common ancestor. And — though it’s less likely when an entire family shares a segment with a match — it’s still possible that a small shared segment isn’t what we want it to be, that is, the result of inheritance from a shared ancestor — a phenomenon called identical by descent or IBD.4 Smaller segments could come to us by random chance, called identical by state or IBS.5
But even the high priority, large segment, match-all-four-siblings matches are not the the only ones we need to look at. Another 366 people match three of the siblings, but not the fourth. And there are some really interesting patterns in that group, where, for example, one sibling out of the three will have two or three times as much shared DNA with the match as the others do. Having three siblings match the person tells us this is someone we should look at seriously, but it’s somebody we might have overlooked — or put low on the priority list — if we hadn’t tested the one sibling who shares a lot of DNA with the match.
Another 589 people match two of the siblings, but not the other two, and we see even more interesting patterns. Sometimes both siblings have very small matches with the match, and we might put that match lower on the priority list. But sometimes both have fairly large segments in common with the match, so even though the other two siblings don’t have enough DNA in common even to show up as a match, this may be someone we want to move higher on the priority list.
And then there are the 567 people — fully one-third of all the matches — who match one sibling, and one sibling only:
• David has 130 matches that he doesn’t share with his brother or his sisters.
• Carol has 154 matches that she doesn’t share with either of her brothers or her sister.
• Mike has 120 matches that he doesn’t share with his brother or his sisters.
• Trisha has 163 matches that she doesn’t share with her brothers or her sister.
Now many of those matches are very low priority: the amount of overall DNA shared by the one sibling with the match is small and the largest segment the two share raises a serious risk that the match is IBS and not IBD.
But that’s not true of all of these matches:
• David has four matches projected to be as close as third cousin and 24 projected to be as close as fourth cousin.
• Carol has seven matches projected to be as close as third cousin and 15 projected to be as close as fourth cousin.
• Mike has 16 matches projected to be as close as third cousin and 24 projected to be as close as fourth cousin.
• Trisha has five matches projected to be as close as third cousin and 18 projected to be as close as fourth cousin.
That’s 32 possible third cousins and 81 possible fourth cousins who — by random chance — match only one of the four siblings.
People who could have the answer to one (or more!) of our brick walls.
People who might have that photo of the great grandparents we’ve never seen.
People who might know where the Family Bible is.
People we could have missed by testing fewer than all four of the siblings.
You see where I’m going with this, right?
Test as many people as you can.
Note: There are some corrections above, thanks to Jim Bartlett who pointed out a mistake in the IBS/IBD alphabet soup references.
- ISOGG Wiki (http://www.isogg.org/wiki), “Autosomal DNA,” rev. 1 Feb 2014. ↩
- ISOGG Wiki (http://www.isogg.org/wiki), “Recombination,” rev. 1 Feb 2014. ↩
- ISOGG Wiki (http://www.isogg.org/wiki), “CentiMorgan,” rev. 1 Feb 2014. ↩
- ISOGG Wiki (http://www.isogg.org/wiki), “Identical By Descent segment,” rev. 1 Feb 2014. ↩
- ISOGG Wiki (http://www.isogg.org/wiki), “Identical by state,” rev. 12 Nov 2013. ↩
I’m looking at one match that fits just this scenario this last week Judy.
I have my very best match outside of immediate family and a 2nd cousin 1X removed that I tested, that did not show up as a match at all for my brother and sister.
The match DID show up too as a match for one of my paternal 1st cousins,-Nancy is her name, but not all for Nancy’s brother and sister. (also my full 1st cousins)
With the match, I have a total 58.38 cM, and the largest segment at 21.18 cM. Expected 2nd-4th cousin range.
Nancy has this match coming in at total 28.81 cM, largest segment at 11.21 cM, but also with a 5.86 cM segment on her X chromosome.
The match for her shows as expected 4th – remote. Not a top match, but gives more clarity to HOW we connect. With the match to a paternal 1st cousin of mine, Nancy’s mother being my dad’s sister.
Communicating with the match has been very helpful.
It is clear the common ancestry is from their paternal side as well, but of course that side of the family is nothing but 1000 questions from dad on back.
(most of my matches are challenges)
NO problem, we are working on clearing those questions up with our results.
I am slowly testing more family members.
$99 is manageable once in a while, but when I win the lottery they won’t know what hit them!!!
Oh I’m rooting for you to hit the lottery, Mark. I can give you a list of my relatives to test then too!
Drat. Now I guess I’d better do all my kids and try and get any of my siblings to cooperate.
Siblings, children, aunts, uncles, cousins, the guys from the corner bar. The usual suspects, Dave!
Thank you Judy. Yet another valuable article – which I have sent to all of my various “Project Forums”. (It replicates my own experience so well, not to mention those of getting my 1st cousins (father’s side, as an example) to test).
For those who are earnestly seeking and who are in my Project groups, when they do the same, they too see the outcome and when they realise the value, nothing stops them in going wider and wider to seek the accuracy that multiple testing can bring to further their knowledge of both their genealogy and their Family History. A “win-win” situation all round, in my opinion. It is still cheaper than golf!
You’re so right about the win-win, Gail!
We tested all 5 siblings .. Found out we are all 1/2 siblings to each other
The matches we thought we had to to one father our mother was married to
Ended up not be a match as our father. He was our mothers 8 th cousin so we kept getting
Distant cousin matches.. Too bad all parents are deceased
Your experience is certainly why all genetic genealogists need to be clear with those they ask to test: DNA tests can tell you things you — or a member of your family — may not want to know. Good luck finding your answers here.
This was SO timely for me!! I just heard Kathleen Brandt speak over the weekend and so now I’m all excited about DNA again. This is a perfect illustration as I try to con, er…sweet-talk more relatives into helping me with this.
We all need that extra encouragement every so often!
Interesting stats, Judy! However, I’m not altogether certain I would automatically give higher priority to matches with all four siblings. It would depend on whether they represent independent lines of evidence (the matching segment is in different places) or if they’re all just saying the same thing over and over (the siblings all match each other on the same segment, because they all inherited the relevant piece of DNA from their parent). I don’t know how hard it would be for you to tease out that distinction, but it would be interesting to know. In any event, I would place the most emphasis on the size of the segment.
I wouldn’t automatically give those four-sibling matches priority, Ann. As noted: “Not all of them are high priority candidates, of course. The smaller the amount of overall DNA that a match has in common with the siblings, and the smaller the size of the largest shared segment, the farther back in time the common ancestor is likely to be — reducing the odds of ever being able to identify that common ancestor. … Smaller segments could come to us by random chance.”
Understood — but would you give more priority to a large segment found in four siblings vs the same size segment found in one sibling?
I probably would give a little more priority to the one in all four siblings, simply because it has proved itself to be “sticky” — coming down unbroken to all of them. That might also have been “sticky” in cousins.
I want to let you know that your blog post is listed in today’s Fab Finds post at http://janasgenealogyandfamilyhistory.blogspot.com/2014/03/follow-friday-fab-finds-for-march-14.html
Have a wonderful weekend!
What an absolutely AMAZING blog post!! I have been wondering about this- they match some but not other type of phenomenon and you explain it so very well. I have tested 2 of my siblings and both my parents and then my son– and I know I have to get on to trying to convince some cousins to test– with each new tester I get a fuller picture!!
Also, on the flip side of this– for OTHER people who are researching their trees- the more people who test, the better results will be found. I’ve had all the spouses of my kids tested- and so far we have seen 2 adoptees find their birth families- and others who have broken through brick walls – plus found information they would have never found if these people had not been tested!
Yes…Test as many people as you can…I AGREE!!!!