GENEALOGY-DNA-L ArchivesArchiver > GENEALOGY-DNA > 2003-07 > 1059065120
Subject: Re: [DNA] from Tony Frudakis re inconsistent siblings on DNAPrint test
Date: Thu, 24 Jul 2003 12:45:20 EDT
Some follow-up from Tony:
Yes, with sib-tests not involving the parents it becomes a likelihood test
rather than an exclusion test (such as parent-child testing, which simply
requires matching alleles between the two and any mismatches like child having
allele 23, 23 and parent having allele 21,15 mean the child is not from the
parent). Such a test can only be done using STRs if one measures small numbers of
markers. It could be done with SNPs but 100 or more would have to be measured.
Uncommon alleles shared in common among two people makes it more likely they
are siblings than if there were no uncommon alleles shared. Full sibs are
expected to share a certain number of alleles, which is more than half sibs,
which in turn is more than two randomly selected people. The likelihood values
produced by the test vary from case to case because they are dependent on how
common the alleles are that two people share, which vary from case to case.
Though the ancestry test and the sibling test are both likelihood based
tests, the important thing to remember is that they use completely different
markers, and of course different algorithms. Virtually never would you find a pair
of people who were determined to be less than full sibs confirm as full sibs
in parent-child testing. In the case of these sisters, the ancestry test was
essentially telling them that there might be a problem with the perception of
full-sibship, and the STR sibship test confirmed it.
In your case, I would not be surprised to see that you are indeed full sibs.
In the case above one sister had almost 30% non-euro admixture and the other
none. That is very different from you and your sister, who both have about
9-14% admixture. In that case, I would suspect the test/program simply missed a
percentage below the threshold of a few percent for both of you (in other
words, you would really be closer to 4% EAS, 5% NAM and your sister 9% EAS 5% NAM
or something similar - you get the idea).
You can post this too if you like.