I've been wondering this myself. I don't really agree with the other comments saying it's impossible.
We do genetic testing on the medical side and that data is kept private. I don't see why a company couldn't offer similar stuff, paid privately, for a more comprehensive suite of tests. You could learn about your risk factors and keep the data private.
On the history/ancestry side, it could pick out known biomarkers to trace back from publicly accessible data. You wouldn't be able to track down exact family trees, but I don't think that was the intent since you're looking for privacy. Instead you might get stuff like "you're 40% Greek, 20% North African"
Such a company would
collect a sample
compare the data against literature
delete the data
It could also allow customers to opt in for more detailed analysis (for those that don't care for privacy) and let them know about the risks. Or it could give an option to share anonymized health data for researchers investigating diseases / risk factors
For the record, ancestry dna is basically a scam. Especially when they give you a percentage score.
Ask yourself what is French. Or English. How much interbreeding has happened across the spectrum? It can’t tell you who you are- there is no genetic encoding for culture.
This makes sense, I don't really know how they come up with those numbers. I feel like there is a realistic risk for harm if we DID try to classify it (ex. If you have X gene, you are Y race). It wouldn't make any sense to begin with, and it would enable arbitrary persecution
I'm more familiar with the inverse, where doctors can provide better care by screening for risks and generic markers that are more common for a particular demographic. That actually helps humanity and is worth studying more
I don't really know how they come up with these numbers
They sample multiple people from a given region of the world and then look at possible genetic similarities between most individuals in that sample.
Then, they collect your genetic data and "match" to all the different signatures they've collected from different regions, and compute a similarity score.
In theory, if they had sufficient samples and the genes were very characteristic, this could work. In practice, any geneticist will be able to point out multiple flaws with this methodology.
There are indeed certain traits that only occur in specific populations... And while someone else totally unrelated could randomly have a similar mutation, it'd be unlikely. But those are rare, and absolutely not something that can be used to say "78% German"
Yeah I’ve always thought when they give those stats “how long ago?”. Where people’s ancestors lived could be quite different during different time periods, that I don’t think can be accurately represented by percentages.
particularly considering the way they establish them is by comparing you to modern genetics in those groups, and maybe a census of how they identified in the mailer. But our genetic pool is as clear as mud; there's a lot of mixing going on between groups;