How precise is DNA paternity testing?
January 28, 2014 7:49 AM   Subscribe

Consider this hypothetical scenario: My father is dead, as are both his parents. There are no samples from any of my direct paternal ancestors. The only source of DNA from my father's side of the family is from three of my father's living siblings. Using those samples, can DNA testing prove (to the satisfaction of the courts) that I am my father's child, and not simply someone probably related to the family somehow?

I've tried to answer this for myself on DNA labs' web sites, and this is exactly the point where they stop saying "Yes! We can prove or disprove paternity based on these samples!" and start talking in weasel words.

Bonus Questions: Would paternity be harder to prove if the lab can only get samples from one or two of those siblings? And would it be useful to have a DNA sample from my mother?

To repeat, this is hypothetical. I'm not really trying to prove my parentage. Just trying to understand what is and isn't possible with DNA testing.
posted by Naberius to Science & Nature (10 answers total) 2 users marked this as a favorite
This is called "Reverse Paternity Testing," if that helps your search. Whether or not it would be accepted in court would depend on why paternity needs to be proven or disproven, which affects what standard of proof is needed.
posted by muddgirl at 7:59 AM on January 28, 2014

Scientist with some genetics and genomics background here. Let me say that weasel words exist because typically one can only make statements that begin with "it depends". Assume that everything I'm about to say begins with "it depends". Let's begin.

First, let's bounce a question back at you. You'd like to classify your hypothetical person as their father's child or "Someone probably related to the family somehow". Please be more specific about that second category. Does that category include cousins, half-siblings, twice-removed relatives, etc? That would make answering your main question a lot clearer.

For your bonus questions: People get a random-ish mix of 50% of their genetic material from mom and 50% from dad. So having mom's sample would be helpful in determining just how much of what we're seeing in the various siblings is actually useful. Also, every sibling will have a somewhat different mix from mom and dad, so the fewer siblings you sample, the harder it will indeed be.

Finally: Is your hypothetical individual male? And are any of their siblings male? Because you get your sex chromosome (Y) solely from dad, so your Y chromosome should be really damn close to your presumptive brother's Y chromosome. That would almost certainly be a lock.
posted by Mercaptan at 8:09 AM on January 28, 2014 [2 favorites]

With the Y chromosome testing (assuming you are male and you have DNA from paternal uncles) it would be easy to say that you are the child of either your father, or of one of those uncles, or of, say, your "grandfather" (or great-grandfather), but difficult to say which one. So you would most likely be using the DNA data in concert with other information (like, if your uncle was four years old when you were born, then you can likely eliminate him from consideration as being your birth father).

Basically, the DNA data alone is useful but it's much more useful when you have other information to use along with it.

On average, siblings share 50% of their non-sex-chromosome DNA. But that's on average. In theory, there could be two siblings with the same parents who have absolutely no DNA overlap, although this would be extremely rare (like, so rare it doesn't really happen). So yeah. It depends!
posted by mskyle at 8:22 AM on January 28, 2014 [1 favorite]

Let's assume my hypothetical person is female, and an only child - it sounds like this is actually a different (harder) puzzle than if they were male?

And to be clear, I'm not trying to prove that she is "Someone probably related to the family somehow." That's "as opposed to" not "or." (I'm assuming DNA analysis would make it clear that she's got a lot more overlap with the alleged aunts and uncles than a random person would have. Am I wrong?)

In other words, could we nail down her specific father beyond the (I'm assuming relatively noncontroversial) statement that she fits somewhere in the family tree but we can't say exactly where.

It sounds like the answer is maybe, maybe not depending on how cooperative the father's siblings are in providing samples and just blind genetic luck. Is that more accurate?
posted by Naberius at 8:43 AM on January 28, 2014

Male children have the exact same Y-chromosome as their father, grandfather, paternal uncles, and brothers (more or less). Female children share an X-chromosome with their father, but may not share one with their brothers, grandfather, or uncles, so to improve your confidence you would need more and more samples.

Also remember that a court of law can and does consider non-DNA evidence in paternity cases.
posted by muddgirl at 8:57 AM on January 28, 2014

Okay, I'm going to say you can determine this if you are male and depending on whether there are other brothers of your putative father.
First, test the Y-chromosome of any of the three brothers of your putative father and compare it to your own. The results will be:
1) You are not related.
2) You are this man's nephew or child. The Y chromosome will not say which.

If the answer is number 1, you are not in this family. If it is number 2, do a regular paternity test of multiple chromosomes, let's say fourteen of them, for each of the three brothers of your putative father. If you share half of the chromosomes of one of them, you are that person's child. If you share much less than half, you are their nephew of all three. The only opportunity to be nephew of all three is to be the son of their absent brother.

Of course, if there are other absent, untested brothers, then you could be that person's child.
posted by dances_with_sneetches at 9:12 AM on January 28, 2014

Extending the above to female. It is trickier. You can not use the Y-chromosome from the three brothers. You could still jump to the second half of the test and test all three brothers for how many matches they have with you. If one is your father, that person will share 50% of the chromosomes. If none of them is your father but you are their nephew, they will average 25% shared chromosomes with you. Furthermore, when one chromosome isn't shared in one person, that chromosome will peek up in another.
The statistics will be complicated, but I would say that you could come out with something like a one in a million chance of error stating that the female child is a nephew to all three and therefore the child of the absent brother.
You could eliminate the possibility of being a child of a sister by mitochondrial DNA testing of any of the brothers or by testing the actual mom, if known.

And, in answer to your other question, it is trickier if you can not get the DNA samples from each of the brothers. You will be left with the question - is the holdout brother the actual father? (You are merely proving to be a niece/nephew of the other two.) The statistic chance of sharing 25% between two brothers will drop down, but will still probably be 1 in 10,000 or more.
posted by dances_with_sneetches at 9:24 AM on January 28, 2014 [1 favorite]

Even if you are male and your father's siblings are male and you do as dances with sneetches suggested... Are you 100%, beyond any possible doubt positive that your father and his siblings share the same father? With Y-chromosomes you may be able to tell if you are related through the Y-chromosome to his siblings, but if you are not that doesn't prove that you are not your father's child, as some hanky panky could have happened a generation earlier.
posted by brainmouse at 9:30 AM on January 28, 2014 [1 favorite]

100% certaintly is impossible even for a straightforward paternity test. That's why labs give confidence ratings and probabilities, which courts will have to interpret based on other facts presented in the case.
posted by muddgirl at 9:47 AM on January 28, 2014

I haven't seen anyone talk about tandem repeats yet. Basically, you have segments of DNA that are repeats of a short sequence. The number of repeats is your allele, and you get one from each parent for each location in your genome (paternity tests can use dozens of loci or more). These loci are highly variable, with many different possible alleles at each locus. If you have a sample of your mother's DNA, you can determine which alleles are from her, and you can compare your other allele to your paternal uncles' alleles - at least one of them will likely have most of the alleles your dad had. You should be able to eliminate the likelihood that your uncles are your bio-dads this way (no perfect matches with your alleles), while building the case that your dad is your dad, because he would share each allele with, on average, half of his brothers. The exact likelihood depends on how many loci you look at, how rare the alleles in question are, and whether you have a DNA sample from your mom. It also gets more complicated if there's any doubt that your dad's brothers are full, bio-related brothers.

If you can check out an intro genetics textbook from the library, it should have a pretty good overview of paternity testing / determining relatedness.
posted by momus_window at 10:49 AM on January 28, 2014 [2 favorites]

« Older Can you show us how to solve a...   |  Any tricks, hacks, tips to get... Newer »
This thread is closed to new comments.