DNA analysis and siblings

I understand that full siblings receive half of their DNA from each parent, and that each sibling’s half does not entirely match the half received by the other siblings. Each parent can have two different versions of each gene, and it’s a coin flip as to which one any sibling will have.

My question is about the ethnicity or geographic origin data provided by companies like 23andMe. Will the percentages match among siblings? Or is it possible that the genes received by one sibling will show different percentages of ethnic background than another sibling?

Back of the envelope, I’d expect to see differences of 5-10% between siblings (like, one might show up as 50% German, while another is only 45%). But it’d depend on how you got those percentages, not just what the percentages are.

For instance, I know a family where the mother was a German immigrant, and the father was a Mexican immigrant. If we assume that the bloodlines of both parents are “pure”, then their kids would all be guaranteed to show up as 50% German, 50% Mexican. But if one of those kids met a kid from another family like theirs, who was also 50% German, 50% Mexican, then their kids would likewise be, by ancestry, 50% German and 50% Mexican… but their genes could theoretically be anywhere from pure Mexican to pure German, on a child-by-child basis (though of course the mixes closer to half would be more likely).

Siblings should match to within the margin of error. The margin of error will be similar to polling errors; instead of sampling a subset of people, they’re sampling a subset of DNA.

If their sampling size is approaching the total* number of DNA snippets and the expected number of DNA snippets in a category is low enough, then you might see true (as opposed to sampling) variation between siblings. If that is captured in their margins of error, you’ll see larger margins in categories with smaller percentages.

*Not all DNA is useful to distinguish people. We share some large percentage of our genome with all life; none of that helps determine where your ancestors lived.

I’ve thought about doing one of those DNA services, and if I do, I’d offer it as a gift to my siblings as well, just to get more confidence in their results.

No, the percentages may not match very much at all with siblings. There’s a woman with a blog who had all of her siblings take the same test and they got different results. Legal Genealogist’s DNA results from her tests. She cites her stuff as well so it should be a good place to start.

Nice link. And that’s why I’d want to see my siblings’ results as well. The margins of error are huge, swamping out the measurements entirely.

Bit of nit-pick, but it’s hard to define genes for being “Mexican” just like it’s hard to define genes for being “American” or “Canadian”, unless the person is Native American. Most Mexicans are Mestizos, meaning they are a mix of European (usually Spanish), Native American and sometimes African and even Asian. Middle Eastern roots are not uncommon, either.

Thanks for the answers, especially Edward The Head. That link, including the replies to the blog post, told me more than I even knew to ask.

True; I was just using that as the example because that’s the family I happen to know personally with immigrant parents from two different nations. But it would have worked a lot better pedagogically with two ethnicities which are much more distinct.

She does a lot on DNA which should provide a lot of information. I don’t follow too much of her stuff, but she does a lot of interesting genealogical work, and a lot of things to do with the law. It gets a bit too much for me, but she does source everything which is why I linked to it.

Here’s some actual numbers for my sister and myself, per 23andme -

Overall, we share 50.8%

European 99.3%98.9%
Northwestern European 97.5% 98.3%
British & Irish 60.6% 63.5%
Scandinavian 5.3% 5.0%
French & German 4.9% 10.3%
Finnish 0.0% 0.2%
Broadly Northwestern European 26.8% 19.2%
Southern European 0.6% 0.0%
Sardinian 0.3% 0.0%
Broadly Southern European 0.3% 0.0%
Broadly European 1.2% 0.6%
South Asian 0.6% 0.7%
Broadly South Asian 0.6% 0.7%
East Asian & Native American 0.0% 0.2%
East Asian 0.0% 0.2%
Yakut 0.0% 0.1%
Mongolian 0.0% < 0.1%
Broadly East Asian & Native American 0.0% 0.0%
Middle Eastern & North African 0.0% < 0.1%
North African 0.0% < 0.1%
Broadly Middle Eastern & North African 0.0% 0.0%
Unassigned < 0.1% 0.1%

I tried to fix the formatting - didn’t work

European, South Asian, East Asian & Native American, and North African are the broad categories

Under European, NW European, and Southern European and Broadly European are subcategories, with sub areas under that.

South Asian, East Asian & Native American, and Middle Eastern & North African are likewise broken down.

There are obvious problems* differentiating “British” from “Scandinavian”, so it’s not too surprising that you’d see that kind of result.

*Much influx of Scandinavians to the British Isles over the centuries. Starting from the Anglo-Saxon invasion, at least. Remember, it was Anlges, Saxons and Jutes. Jutes are basically Danish.

Think of it this way -
There are 23 pairs, 46 chromosomes in the human genome.
You get one of each of those pairs from each of your parents’ pairs.
So think of the selection of your chromosomes as flipping a coin - heads you get a grandma’s chromosome, tails you get grandpa’s -from each of mom and dad.
Flip a coin 46 times in a row - 1-23 for inherit from dad, 24-46 for mom.
Now do the same for a sibling.
What are the odds that when you got a Heads for #1, they got a Tails (different chromosome)? or Heads (same chromosome)?
After 46 flips, how many flips agree between the two runs?
Generally, the odds that around 50% agree is pretty strong.

That’s leaving out the important process of Chromosomal Crossover. That’s what makes it likely that the number isn’t precisely 50% (ignoring the sex chromosomes, which can introduce more variation). And this multiplies throughout the generations such that at 7 or 8 generations, you start having a finite probability of getting 0 genes from one of your ancestors.

Chromosome crossover actually means that you’re likely to be closer to 50% to your siblings, and less likely that you’ll get zero genes from any given ancestor.

Huh, this has been an interesting thread to me.
My parents did a dna test through 23andme.
As to how the categories of origin are arrived at(European, Northern European, British etc) seems problematic and vague. What’s a “British” historically, or a Northern European?
Mom’s family (both sides) is well documented to be all German. We even know who our family are in “The Old Country” and some of us have visited our cousins. Imagine my mom’s surprise when her results came back 35% british. (seriously, wtf is a british?)

Perhaps some people have the mistaken idea that, sometime in the not too distant past, there were European countries with genetically “pure” populations and these testing services are figuring out how much your ancestry derives from these pure populations. Of course, there never were such populations, and the tests aren’t comparing your genes to past populations anyway. They are comparing your genes to present populations, even if they try to screen for people who “know” that their ancestors have lived in a certain area for generations.

And you’re absolutely right about what does it mean to be British, genetically. Especially since most of us should have learned how those islands have been invaded over and over again by various continental populations.

My mom’s family is over half German (all of Gramma’s known ancestry was from Germany, plus some of Grampap’s), and yet, when one of my aunts got tested recently, it came back as no German at all. Several of her siblings are now getting their own tests, both from that company and from others, to try to figure out what the heck happened there.

Comparing two non-identical siblings, recombination means each within each chromosome there are 3 or 4 alternating stretches of shared/unshared DNA. Statistically, I don’t think that increases the variance of total shared DNA, it reduces it further (i.e. makes it closer to 50%).

As for the second thing… are you talking about looking back to a generation where (by chance) none of your DNA came from one of the ancestors in that generation? Again, recombination has the opposite effect of what you suggest, since it divides the genome into a larger number of segments where ancestry alternates. It pushes that generation further into the past.

ETA: ninjaed by Chronos - well, not exactly ninjaed, didn’t notice his comment.

:smack: Oy, brain fart on my part. Yes, the recombination or cross-over makes it virtually impossible for you to get 0 DNA from a grandparent. It is important in calculating how far back you need to go before you have a non-zero chance of getting no DNA from a given ancestor, but it pushes that point further back in time.