Yes, DNA can usually tell the difference
It’s one of the most common questions people have about DNA testing. Just this past week two different people wrote in to The Legal Genealogist with some variation on the theme.
Can a basic genealogical DNA test tell the difference between half-siblings and full siblings?
Most of the time — as in one of the questions this week — the questioner’s parents have passed on but brothers and/or sisters remains, and there are still niggling doubts about whether they all share the same set of parents.
Could they be half-siblings — sharing only one parent — instead of full siblings, sharing both mother and father? And can a less-than-$100-buy-the-kit-and-wait-for-the-results DNA test really tell the difference?
The answer, of course, is usually yes, it can tell the difference, as long as the test results are combined with all the facts and properly understood.
Now… let’s be clear here: we’re talking about the autosomal DNA test here, and not YDNA testing or mitochondrial DNA (mtDNA) testing. Autosomal DNA is the kind we all inherit from both of our parents and that’s used, in genealogy, primarily to find cousins to work with collaboratively.1 YDNA is the kind only men have that’s inherited only from their fathers and so is used to trace the direct paternal line.2 And mtDNA is the kind we all have but that’s inherited only from our mothers and so is used to trace the direct maternal line.3
Here’s the deal: with autosomal DNA, we all inherit 50% of our DNA from our fathers and 50% from our mothers. Because of a random jumbling of the DNA that occurs every single time a child is produced — a process called recombination4 — brothers and sisters won’t share the same exact DNA. One sister might have gotten a bit more from the maternal grandmother’s line than the others; one brother might have gotten a bit more from the paternal grandmother’s line than the others.
But on average full siblings will share about 50% of their autosomal DNA.
Now… since half-siblings are only inheriting autosomal DNA from one shared parent and not two, the amount of shared DNA they’ll receive is cut in half: to 25%.
And that difference usually shows up fairly dramatically in the standard DNA tests we take for genealogy when we look at the number of centiMorgans (cM for short) that we share with a sibling. That’s a unit of measurement in DNA that helps determine genetic distance — in other words, just how closely related two people are.5
The higher the number of cM shared with a sibling, the closer the genetic distance — or, in plain language, the more likely it is that you’re full siblings and not half.
Now the charts we used to use in genealogy tell us we should expect 2550 cM in common with a full sibling and 1700 cM in common with a half sibling.6 But more recent research has shown us there’s a range we can expect to see.
The Shared CentiMorgan Project, pioneered by Blaine Bettinger, tells us that the average for full siblings is 2629 cM, and the range is reported from a low of 2209 to a high of 3384. It also tells us that anything above about 2900 is a real outlier and not likely to be a correct interpretation of the data.
For half siblings, the average is 1783 cM, the range from 1317 to 2312 but anything above about 2150 is a real outlier and not likely to be a correct interpretation of the data.
That’s shown in the histograms on the Shared CentiMorgan Project: the charts you can see there that show you where most of the results show up. The results out at the ends are the outliers.
You can see this represented in terms of the statistical probabilities — the odds that two people are full or half siblings — in the DNAPainter tool called The Shared cM Project 3.0 tool v4. Plug in the total cM in common in the box at the top, and the tool will give you the odds: at 1700 cM, the odds are 100% that the two people are half siblings; at 2620 cM, the odds are 100% in favor of full siblings.
My own half sibling, for example, has a range of 1622-1975 cM with his half-brothers and half-sisters. The percentages tool gives us 100% odds for half sibling on that entire range. And my range with my full siblings is from 2566-2858 cM, with the odds for full sibling from 99.71% to 100%.
Where it gets dicey is starting around the 2000 cM level — that’s where there is some small amount of evidence that a match could be either full or half sibling. At 2250 cM, the odds are pretty close to 50-50 either way. And at that point, of course, you may very well need more than an over-the-counter DNA test to answer the question.
The results can also be skewed by issues like one parent being an identical twin (say two identical twin brothers each had a child by the same woman) or by incest.
Bottom line: yes, usually, the test data tells you whether you’re full or half siblings, as long as you look at all the data, and understand the test results fully.
Oh… and remember… there are real ethical issues involved in DNA testing, and we need to be prepared for all of them before we ask anybody — especially a maybe-full-maybe-half sibling — to test.
SOURCES
- ISOGG Wiki (http://www.isogg.org/wiki), “Autosomal DNA,” rev. 15 Sep 2018. ↩
- Ibid., “Y chromosome DNA tests,” rev. 4 Dec 2016. ↩
- Ibid., “Mitochondrial DNA tests,” rev. 1 Oct 2018. ↩
- Ibid., “Recombination,” rev. 2 Dec 2017. ↩
- Ibid., “centiMorgan,” rev. 21 May 2017. ↩
- See ibid., “Autosomal DNA statistics,” rev. 3 Sep 2018. ↩
Nicely written, Judy. Of course there are some other possible scenarios adding complexity. The assumption is that the individuals in your sample are really siblings at all – a half-sibling is in the same “shared cM range” as niece/nephew//aunt/uncle. And, particularly in earlier decades, for a child born out of wedlock to be brought up by the parents of the girl and told that he/she is a younger sibling.
I also encountered an uncomfortable situation among distant cousins where father/daughter incest was involved. The father is long deceased but the daughter is still alive – apparently having had two children by her father (both adopted at birth) before the girl was able to break free and establish her own family. I worked with 2nd+ cousins (including a son of the adopted children) who matched significantly more closely than expected. Only because I had a detailed traditional tree was I able to puzzle out the likely connections, all of which were eventually confirmed in interviews with family members having close knowledge. Unfortunately Blaine’s tables don’t consider such issues (not surprisingly, of course). I’d be interested in learning what you would expect to see as shared cM numbers when your 1C1R proves also to be your 1C (i.e., your great-grandfather is your grandfather). How far outside the expected range for 1C will it be? I couldn’t figure it out logically but the acrual valuse in this case were about 950 cM.
Well, here’s one that is an outlier. Three brothers share 2239, 1770 and 1681 with a half-sister. Genealogical research shows her to be the daughter of their father. The three brothers all share over 2000cMs with each other
Bonnie, that’s not that much of an outlier. The odds are almost 50-50 for 2239 cM (53.71% in favor of a half-sibling, and 46.29% in favor of a full sibling).
There are always outliers. That is why you need to look and see if there are any fully identical regions (areas where there is the same DNA from each parent known as FIRs) when the number of cMs does not give you a definitive answer. See https://blog.kittycooper.com/2016/04/full-versus-half-sibling-dna-matches/
One other tiebreaking technique you can use: A half sibling should only be related to one parent. So if you have an aunt or uncle or close cousins on your other parent’s side, if your sibling is not closely related to all of them, then it must be a half-sibling.
I am female and I share 1,033 CM’s with a man convinced he is my half brother. He believes we share the same father. My full sister also tested and shares only 773 CM’s with him and 2,355 with me. His adoption papers describe (without naming) my fathers situation at the time (married man with 3 children). This man’s given name at birth was my fathers name, with a slight misspelling of my last name. Is there any possibility of him being a half brother with such low shared CM’s?
It is unlikely. If you enter those numbers at the Shared cM Project 3.0 tool v4 at DNA Painter, the odds are very high that the relationship is different: Great-Grandparent, Great-Aunt / Uncle, Half Aunt / Uncle, 1C, Half Niece / Nephew, Great-Niece / Nephew, or Great-Grandchild. Some of those can likely be eliminated based solely on the ages involved.
1033cM/15% shared DNA with me, and 771cM/11% shared DNA with me. Ancestry states 1st-2nd cousins. Can these be half siblings to me? My father is unknown. These matches would be from his side.
Nope, too low for half-siblings (see https://dnapainter.com/tools/sharedcmv4) but well within the range for first cousin of half-niece/nephew.
How can I have more dna cm with one aunt and not the same with the other aunt from my fathers side
Easy. Look at it this way.
Dad has 50% Mom and 50% Dad. But say that’s 20% Grandmom and 30% Granddad on his Dad’s side, and 30% Grandma and 20% Grandpa from his mom’s side. Aunt #1 got the same: 20% Grandmom and 30% Granddad on her Dad’s side, and 30% Grandma and 20% Grandpa from her mom’s side. But Aunt #2 could have gotten 30% Grandmom and 20% Granddad on her Dad’s side, and 20% Grandma and 30% Grandpa from her mom’s side.
Since you could only get what your Dad could give you, you could have a lot more Granddad and Grandma in common with Aunt #1 than you could with Aunt #2.
I don’t share the same amount of DNA, measured in cM, with each of my mother’s siblings who’ve tested:
Uncle #1 1965
Uncle #2 1871
Uncle #3 1809
Aunt #1 1868
Aunt #2 1824
I don’t even share the same amount of DNA with my own full blood siblings:
Sister 1 2961
Sister 2 2760
Brother 1 3075
Brother 2 2760