I have been meaning to make a thread about this for a while now. And acutally I am glad I waited. We are now doing population genetics right now, so I actually have a better handle on why these are a waste of money.
Ok so first things first. One must always look at the source of the information. The companies performing these tests make incredible claims, but.. An example of one of the companies doing these.. Mars 'Wisdom Pannel' is one that taken by your vet and sent off to the company. This is the same company that brings us M&Ms pedigree dog food and Uncle Ben's instant rice. Intersting no?
Ok anyway off to the science...
When they take a sample of a dog, or a human, in order to run a profile, they only look at a small number of alleles (locations on genes) They look at non coding regions of the DNA. So the alleles they use as markers do not code for blue eyes, long hair, brown spots, etc etc.
CODIS for example uses 13 loci to profile people. In lab this year we are profiling ourselves (and I get to profile my son as I got a great DNA sample from a tooth the dentist extracted...) So once you have mapped your alleles at these 13 loci, then you compare them to a database.
This is where the problem arises. Your results are only as good as your database. This is also where it gets tricky with humans in the court system too.
How you 'determine' if your suspect/dog is from a subgroup/breed is to calculate the frequency of those alleles in different databases. So if your alleles are more common in African Americans, then they assume you are African American. (and that can be wrong too) But lets say you only have a good data base of Caucasians, African Americans, native Americans, and Hispanics. You get your allele frequencies back and you compare them and you have most in common with Hispanics...what if you are Polynesian? You can't show up as that, as there is no database (this is all fictional, I am sure there is a Polynesian database somewhere)
Now this issues with the dog DNA tests is that they are only comparing a small number of breeds and the DNA is almost all from US databases. So if your dog has any foreign ancestry that can mess up the results. If your dog has a breed that is not represented in the database then it will mess up the results. Also many breeds share common ancestry, this too can skew results.
When I was researching this on the net there was a news show that covered this. All the news anchors had their dog's DNA tested. It was interesting how the one anchor with a beautiful pure bred golden retrieve was told his dog's DNA showed he was a mix (forget what the mix was, but it was pretty funny) he was obviously upset.
The other main issue I have is that the general public puts so much 'faith' in DNA results (thanx CSI) but DNA results are not that cut and dry. Esp when we are talking DNA + statistics. I get very leery when people start acting like the DNA tests are accurate (this is not to bad on this board) As soon as this gets wide public acceptance, BSL will take on a whole new meaning. That is a scary scary thought.