[tharkun__]

And in any case, you may not really get any conclusive data back anyhow.

CBC Marketplace sent a _human_ DNA sample to various dog DNA testing places and from some of them got an actual dog breed answer back (instead of the expected "error, not dog DNA").

https://www.cbc.ca/news/business/marketplace-dog-dna-test-1....

So if you send "wolf" to a dog DNA testing place, how accurate would that really be for "wolf" vs. "dog" when all they expect to find is "dog" (and "wolf" is very close to "dog" given overall ancestry)?

[_yb2s]

Knowing how these services work that is not surprising- most only use a microarray that can only identify the approximate relative abundance of specific small fragments of DNA, it does not read the DNA sequence directly. They then probably have a very primitive analysis that assumes you have dog dna, and returns the same breed as whatever reference sample it is closest to. They almost certainly would have no reference data on a highly endangered African wolf.

[tantalor]

Sure but do they not check whether a sample is contaminated? That would seem like a step you can't skip.

[tharkun__]

For human vs. dog I would definitely expect that all of them would be able to tell (given that _some_ of them definitely were able to tell).

For wolf vs. dog I'd not be so certain even for the reputable ones. But then again I'm not a geneticist, so I can't tell you how easy it is to tell their version of "is this C# or Java byte code" :grin:

[_yb2s]

They don’t get the DNA sequence with these cheap microarray based services. However a human or something just as directly related to a dog would give weird mostly nonsense outputs- even if they could not tell what species the sample is actually from, it would be clearly not a dog. They should detect and reject those samples, but I am not surprised some of the services didn’t bother to implement that, as it requires more work.

This is most likely what they are using: https://www.thermofisher.com/order/catalog/product/550869/ta...