blutoski
Penultimate Amazing
- Joined
- Jan 10, 2006
- Messages
- 12,454
Thanks for the snippet
Weirdly, it popped up for me this time, although the last two times it was he heart-cell butterfly. Makes me wonder what kind of rotating algorithm they're using.
Regarding the consumer DNA testing... The cynical part of me wonders how long it will be before they start selling those data sets for analytic purposes. There could be a lot of power in having a massive set of DNA data to mine. There could also be a lot of big-brother kind of risk...
ETA: I've been kind of skeptical about the DNA testing for ancestry anyway. A friend used one for her mutt of a dog, and it came back with a breed list that couldn't possibly produce a dog that looked like her... I mean, it's horribly unlikely that chihuahua, poodle, pit-bull, and terrier is going to produce a mix that looks a lot like a whippet. Also, my Chinese friend, of Chinese descent is only 97% Chinese - the other 3% is apparently Irish somehow. And my friend with Mexican parentage is apparently 25% Japanese!
Entertaining, yes. Accurate... the analyst in me says not even a little bit. I'd guess they've got an R-squared of something like 40%![]()
I think the complication is that they're testing alleles against current geography, and trying to translate % probability of X descent into % X descent.
Just a more numerical example for illustration, if the allele is 100% of Irish and 100% of Chinese, there's a 50% chance that this specimen has Irish relatives and 50% chance that this specimen has Chinese relatives (I know, I know... plus or minus for adjustment based on relative population size). This is not the same as being 50% Irish and 50% Chinese.
So this is where family history comes in. It can 'collapse' the probabilities into real information. In Warren's case, the family history she cites is very likely true based on its ordinariness in that region and population. An advanced (per earlier discussions, post-research, validated &c) DNA test will not likely provide much more information than we already have.