Before I put this canine DNA odyssey to doggy bed, I wanted to share some, uh, insights from the guy who co-developed the technology used in the Wisdom Panel Insights test, Leonid Kruglyak of Princeton University. My questions in italics:
According to the Wisdom Panel DNA test, my dog is 50% Rottweiler and 25% Chow Chow. Presumably this means that he carries a critical mass of alleles associated with these breeds. Is that correct?
The results are actually more specific than overall fraction of SNPs with alleles consistent with those two breeds. They mean that the chromosomal patterns of alleles are such that they are best fit by one parent being a pure-bred Rottweiler, one grandparent being a pure-bred Chow Chow, and the other grandparent on that side being a very mixed-breed dog.
How does the reliability of DNA-based dog breed assignment compare with commercially available genetic ancestry testing for sub-continental origins in humans (e.g., “West African,” “Northern European,” “East Asian,” etc.)?
I’m not aware of solid accuracy numbers for either type of testing.
I think in both cases a lot depends on the actual ancestry of the dog
or the person — some signatures are harder to pick up than others.
DNA quality also varies in cheek swabs and this contributes to the
accuracy of any given test.
My dog is ostensibly 25% generic “mixed breed.” Is this because some fraction of his alleles are too common to allow for specific assignment? Or they are not found in the Wisdom Panel database? Or something else?
I think this means that the observed genotypes are not consistent
with any one breed contributing at the great-grandparent level (or
closer). Further back in the ancestry, you get too few SNPs to
reliably determine a breed contribution to the tested dog. (Note that
generally SNP alleles are not breed-specific but vary in frequency
among breeds, so determining a breed contribution is a probabilistic
calculation across multiple SNPs, just as it is in human ancestry
This work, unless otherwise expressly stated, is licensed under a Creative Commons Attribution-NonCommercial 3.0 Unported License.