For those not familiar with the issues of trust and intellectual property and genetic testing in Native communities:
Havasupia v University of Arizona Board of Regents
What needs to be understood about these tests is the quality of the information extracted from the data is dependent to a large degree on the size of the database used. The clinical side of these tests is relatively easy.
The human genome is huge -- about 3 billion nucleotides. But ca. 99.5% of our DNA is pretty much just like anyone else's. Less than half a precent of the base pairs differ between individuals. So, only alleles (sections of DNA that encode a particular protein) that are polymorphic -- have usually one nucleotide base pair difference -- are useful in this type of work. And roughly 1% of these have sufficiently high frequency that 1% or more of the population has a particular variation. Then these variations need to show stability through the process of procreation and over the span of generations. The correlation of an allele to disease, phenotype variation, or population distribution is teased out through complex statistical analysis.
While the percentages of variations that are present are small, the number of alleles is huge. So sample sizes used to find and track variations must also be huge. A lot of the genetic breakthroughs are functions, not so much of improved clinical chemistry, but of ever larger numbers of genes mapped and individuals sampled. Below is a paper from Nature that will give you some idea of the kinds of techniques used. (Don't be alarmed if you don't understand any of it. This information is so specialized only those in the field understand such papers. Personally, I just read 'em and think wow this is what happens in you stay awake during statistics and don't you just love your acronyms.)
Nature: Human Haplotype Map Project, Phase I
So, when you buy these tests, you're not just buying what markers on which genes, but you're buying computer algorithms used to tease out trends.
What I have never really understood is the urge to hunt ancestors among the base pairs (disease is another matter). I've never really bought into a blood connection to culture, except as your kin shape your culture. With epigenetics is pointing to ancestral environmental experience affecting the expression of inherited genes and neuropsychology to language shaping cognitive structures, we are far more "nurture" than "nature" I think. The only thing I find in DNA is an awesome testimony to unfathomable scope and complexity of the Creator's handiwork.