5

I have two people who have both autosomal DNA tested. I have both of their match lists which contain all the people each of them matches to.

For each person that each of them match to, I have:

  1. The number of segments they match on,
  2. The total centimorgans (cM) and SNPs that they match on, and
  3. The centimorgans and SNPs of their largest match.

If these two people match each other with approximately 3384 cM (at Family Tree DNA, other companies be slightly different), then we know that they are either parent-child, identical twins, or the same person who tested twice.

We can easily figure out if they are identical twins or the same person who tested twice since almost all the matches between the two will be the same. If the match lists are quite different, then we know we have a parent and child.

My question is: Given the match data for both people, is there any way we can tell who is the parent and who is the child?

  • Huh? Wouldn't the kits of the same person who tested twice or the kits of identical twins share twice the cMs of a parent-child match? – Jan Murphy Jan 8 '18 at 1:47
  • 2
    @Jan, yes they do. They are fully identical and share 3384 cM on both chromosomes of each paternal and maternal pair. In total that's 6768 cM. But autosomal DNA tests cannot differentiate the chromosomes in each pair. So any matches on both chromosomes are reported as if they match on just one and get reported as 3384 cM, the same as parent-child. – lkessler Jan 8 '18 at 2:12
  • 1
    I find that identifying duplicate test kits or identical twins is easy via a GEDmatch.com feature: a 'One-to-one' with the 'Graphics Only' option will display match bars for each chromosome. These bars are colored green where there are stretches where the base pairs on both chromosomes match. For identical twins or duplicate kits they will show essentially completely green bars. (It also helps with identifying matches between full siblings since they will usually have substantial stretches of green.) – RobertShaw Jan 8 '18 at 18:18
  • This seems like an odd situation. You have access to DNA results from two people, but you don't know their ages? Obviously, knowing the ages would resolve parent/child. – Jamie Cox Jan 11 '18 at 15:04
3

Just by comparing the two tests, I don't think you can distinguish parent from child. The match is symmetrical. However, if you look at other matches in their lists, you can. Suppose there is, say, a 3rd cousin level match which matches both the parent and child. That person will probably match the parent with more centimorgans (cM) than they match the child because the child usually only inherits part of the matching DNA.

You have to be a little careful with a single match, because that person may coincidentally be related through the child's other parent. But, if you find several somewhat distant matches which match both A and B, but which match A more closely, then A is the parent.

  • Why do you need other any more match lists? The third person should appear in both the parent and child's match list. Shouldn't having just their two match lists provide you with the matching cM to the 3rd cousin? – lkessler Jan 11 '18 at 0:30
  • 1
    I think we are in agreement. No, you don't need any more match lists. Yes, you need to look at other matches besides the parent and child. I will clarify my answer on that. – Jamie Cox Jan 11 '18 at 14:58
  • I think you've got the right idea (so I'm +1 for you), but there appears to be quite a bit of randomness to the matches, and I need it to work even in an endogamous population. For example, some actual data I have: The father matches to 12,613 people. The son matches to 12,337 people. They both match in common to 8,301 people. Of those, the son has larger matches to more people in common than the father does: 4,343 to 3,958. Of the matches in common, the son's average match length is 83.0 cM compared to the father's average of 81.3 cM. – lkessler Jan 12 '18 at 0:24
  • Part of the reason in this endogamous population might be that the son matches to a lot of people on his mother's side who his father is distantly related to with small cM matches. Also as you say, the son seems to often be related to his father's relatives through his mother as well, which makes quite a number of his matches to his father's relatives more than his father's match amount. So might you have any further ideas about what can be done? – lkessler Jan 12 '18 at 0:30
  • I think my answer would work in most cases, but if you're dealing with that much endogamy, all bets are off. It sounds like you are trying to program this so it can be determined automatically by an algorithm. Sounds like an interesting project -- good luck. I think that mass statistics (e.g. 12,000 for A, 8000 for B) are less likely to help than careful analysis of a smaller number of closer matches. Any additional data you have may help. Do you know in general the sex of the two? Do you have any X DNA information? Segment information? – Jamie Cox Jan 12 '18 at 1:00

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.