Family and DNA

Factual Questions
1

I have never really studied genetics. I don’t have the right terminology for this. Apologies in advance.

First, I do know that all humans have the vast majority of their DNA in common. I assume there’s some term that refers to the rest – the part of DNA that varies from person to person. For simplicity’s sake, let’s just ignore the in-common part, and refer to the “varies by person“ part as just DNA.

As I understand it, a person gets half their DNA from each parent. If you did a 23-and-Me analysis on a parent and child, there would be a 50% match. (Of course, this assumes that the parents are not related to each other. If the parents are, say, first cousins or a Chinatown “She’s my mother! She’s my sister!” situation, there would likely be a greater than 50% match, no?)

(OK, thinking about this, I suddenly realized maybe that 50% business is always wrong. Parents share traits with each other. If mom and dad both have blue eyes, their offspring will have blue eyes, and rather than half of the child’s eye-color gene matching each parent, all of it will match each parent. Anyway, let’s ignore that and just go with 50%.)

So how much DNA do siblings have in common? My guess is that it should average 50%, but would probably be less or more than that.

Suppose Mary and John have a child they name Lee. We could divide Mary’s DNA into 2 groups: went to Lee (we’ll call that A), and didn’t go to Lee (B). Similarly John has DNA that went to Lee (C) and didn’t go to Lee (D). If Mary and John have another child (Chris), we’d assume that Chris’s DNA from Mary would be half A and half B, and from John would be half C and half D, so that Lee and Chris would share 50% of their DNA.
But that’s only an assumption. Isn’t it at least theoretically possible for Chris’s DNA to be all A and C (and therefore be basically a clone of Lee) or all B and D (and therefore genetically unrelated)? I realize the chance is miniscule, but isn’t the chance that Lee and Chris share exactly 50% of their DNA also miniscule?

I got to thinking about this due to a letter in Ask Amy this morning. It was about family turmoil being caused by genetic testing indicating that first cousins were in fact half-siblings. I’m thinking that in theory half-siblings should share 25% of their DNA, and first cousins should share 12.5%, but I’m also thinking it’s much more variable so that first cousins would be misidentified as half-siblings more often than people think. So how often would these misidentifications be likely to occur?

Also, do the DNA testing facilities have any prior knowledge about the relationship between the people whose DNA they are analyzing? Suppose Mary has a sister named Alice and John has a brother named Tom. If Alice and Tom have children, won’t those children show as more closely related to Lee and Chris than just first cousins?

TL:DR How much DNA is (Theoretically) shared by various family members?

2

Full non-twin* siblings should have on average close to 25% relation to each other, with the theoretical range running from 0% to 100%, with the likelihood of either extreme happening profoundly close to zero. This is because while each child gets half of their DNA from each parent, each child gets a random mix.

(* Identical twins can be extremely close to 100% identical for fairly obvious reasons. And in theory polar body twins can exist that are close to 0% identical.)

3

Here’s a helpful chart

Double cousins (the children of Mary and John and Tom and Alice are double cousins) are more closely related than ordinary cousins - an ordinary pair of first cousins share 12.5% of their genes, while double cousins share 25%, and are thus as closely related as full siblings.

4

Looks like I’m wrong about the average for full siblings as written above. I was thinking the halving twice instead of once. (But it still is a theoretical range of nearly 0% to 100%.)

5

Yes.

A quick note: I was very disappointed when I left high school and went to university, and learned about genetics again. There are multiple genes involved in eye color, hair color, etc. So brown eyes aren’t necessarily dominant to blue eyes, etc. That’s just stuff they tell high school students to keep things simple. (At least where I live.)

If you leave out the X and Y chromosomes, this is the right average for full siblings. (They’re also leaving out mitochondrial DNA, which is tracked through the maternal line only.)

Yes, although mathematically that isn’t at all likely to happen.

I doubt very often. Two half-siblings might only share 22.5% of their DNA with each other, or 27%, or something like that. That’s far more DNA in common than first cousins would have so there’s almost always going to be a gap. (Note that double cousins share 25% of their DNA with each other, but that’s a pretty rare marital setup in the west.)

I don’t think they will have that info unless you tell them. If you went to 23 and Me, you might find you have a cousin (or some other relation) you didn’t know about, which might even be the point. Now if your cousin was on the site, but your DNA match shows about 25% relation, then you might have questions. (Such sites have been used to identify criminals, as in DNA testing showed the perpetrator was a first cousin or thereabouts with these three or four people, so the suspect tool just got a whole lot smaller. The police might have had no idea who the suspect was before they did the testing.)

By contrast, if you’re on the Montel Williams Show, and are hoping he shouts “you are not the father!”, presumably the testers have some notion of how two people are officially related.

6

Oops. I should have read more carefully - full siblings are 50%, so double cousins aren’t as close as full siblings, but closer than ordinary cousins

7

Can you explain what you mean here? Because I thought that full siblings would have on average 50%, and every source I know says so, but you sound so certain that maybe I just misread you.

I don’t see how a full sibling could share no DNA with their sibling. They got 50% of the genes, or sometimes 49% or 51%, but it’s not like the egg or sperm ever gets completely ignored. Is it?

Phenotypes - the way people appear in terms of the way they look and the genes they express (usually with regard to diseases) - are a different matter. You can appear to be a clone of your father or mother, but you aren’t really.

8

See this.

10

Imagine a simple organism with 4 pairs of chromosomes (because I don’t want to list 23) and ignore crossing-over. Look at one parent. That parent has chromosomes:

1a 1b
2a 2b
3a 3b
4a 4b

Sibling 1 inherits from that parent:

1a
2b
3b
4a

Sibling 2 inherits from that parent:

1b
2a
3a
4b

Therefore the two full siblings share zero DNA from that parent. Assume the same happens from the other parent. The full siblings would share zero DNA.

11

OK. So you got it wrong about siblings. And I just got round to checking about polar body twins, and actually they don’t share 0% DNA either. Where are you getting this from?

12

If they existed, they very much could theoretically share zero DNA. It is profoundly unlikely, but that is the theoretical floor.

13

Sorry, we cross-posted. I couldn’t edit so deleted and re-posted.

I don’t think that’s the way it works. You don’t just get a random number of chromosomes from your parents. Mitosis just doesn’t work like that. But I’m not an expert so hopefully one will chime in.

14

Here it is in much, much more detail.

https://genetics.thetech.org/ask-a-geneticist/siblings-are-around-fifty-percent-related

15

Good thing that I didn’t claim that, then.

16

Both of my parents had green eyes, but my brother had our paternal grandmother’s blue eyes.

17

If Ron and Tom are identical twins and they both marry and have children their offspring are genetically half-siblings even if legally cousins. If they marry Chris and Trish who are also a pair of identical twins their children will genetically be full siblings.

But people usually know when identical twins are involved.

18

While two siblings being exactly 50% related is highly unlikely, it’s still far, far more likely than them being exactly 0% or exactly 100% related.

19

Here as an example for siblings, from a 2006 paper. On 4,401 sibling pairs, they found the mean relationship was 0.498, with a standard deviation of 0.036, and a range of 0.374–0.617.

To put that into numbers that may be easier to visualize, about 80% of pairs have a relationship between 0.45 and 0.55, and something like 99% have a relationship between 0.40 and 0.60.

Additionally, it’s not just a question of being a mathematical probability of being 0% or 100% related, but also a question of biological possibility. Crossing over is a necessary step in human (all mammalian?) meiosis. Genes will shuffle. In order to get two non-twin siblings to have 100% the same genes, the cross over events would have to occur at the exact same location in both pairs of gametes.

The probability of being exactly 0.500 is more a question of measurement accuracy than probability. Different levels of precision and rounding will change the probability 0.5, 0.50, and 0.500 will all have different probabilities. 0.50 will include everyone from 0.495 to 0.504, for example.

20

I’m gratified to see that my suppositions were pretty much correct. Human genetics is every bit as complex as I thought.

On a more lighthearted note:

(cue music)

Meet Cathy, who’s lived most everywhere,
From Zanzibar to Berkeley Square
But Patty’s only seen the sights
A girl can see from Brooklyn Heights -
What a crazy pair!

But they’re cousins,
Identical cousins all the way.
One pair of matching bookends,
Different as night and day.

Cathy’s and Patty’s fathers were identical twins. Even so, is there any chance at all that the cousins would be identical?

Assuming that their mothers were unrelated to each other, I’m guessing No. But I suppose it’s possible that Natalie (Patty’s mother) looked a lot like Cathy’s mother. After all, isn’t there some evidence that identical twins tend to be attracted to the same type?

21

This chart shows empirical data on shared DNA for different relationships. You can get a good impression of the % shared by dividing everything by the amount shared between parent/child (3485 cM), but different companies operate with different total amounts.

The underlying reports also contain histograms for the collected data for various relationships.

And of course the “real number” is “above 99%” and what this is comparing is “patterns in the shortest segments we can (somewhat)reliably tell apart and presume came from a common ancestor”.

The different companies differ on where they draw the “somewhat reliably” line. Some err on the side of reducing false positives, some are more afraid of false negatives.