As it happens, my day job involves quite a bit of this for tree breeding.
these systems compared your genetic markers to those common in various population groups
We do exactly as you have suggested, both for identification of provenance (tracing back to specific population groups or “races”) and identification of specific parents, grandparents etc etc.
For identification of provenance you need few markers, as genotyping groups of populations will produce reliably fixed and diverse results at certain points. For humans, these informative points are well-known and publicly available in many research papers.
For identification of specific parents, grandparents etc you need quite a few markers (which increases cost for testing and analysis time for data crunching) and it can be tough to reliably determine with very high accuracy. We typically use 3000 markers and produce a likelihood, which is good enough for tree breeding purposes. I haven’t spent much looking at the specifics of the AncestryDNA test. But it appears to be a simple provenance analysis. Looking at the markers you have, seeing which ones match to ethnic origins.
This is why the AncestoryDNA test is $149, some are as low as $89, and a paternal DNA test is around $300 for a cheap one.I’m not sure how many markers they screen for. You’ll see about $500+ for a court approved one.
We charge ~$30 per test for volume customers.
Edit: just scrolled down on the AncestoryDNA test. They claim:
Your DNA will be analyzed at more than 700,000 genetic markers.
Which means they’re probably using a chip-based solution with fairly low coverage (they don’t read and re-reach each marker point to make sure it’s producing a highly accurate result for each individual marker). There wouldn’t need to be high accuracy for each individual marker to make a good estimation of provenance.
As an aside:
We use a sequencing based genotyping system. Good for flexibility and cost. We read & re-read each of the 3000 markers we test for about 25-30 times. We need high accuracy when we’re using markers to make predictions about a tree’s potential performance.