A bit of a disappointing sale at 23andMe today. Details on my blog: http://www.yourgeneticgenealogist.com/2011/11/23andme-cyber-monday-discount-today.html
"What's the margin of error for 23andMe's ancestry painting?"
I think it depends on which population you belong to.
For people of Northwestern European ancestry, it seems be very accurate. According to 23andme: "we're able to say with confidence that a person with an all-European Ancestry Painting (actually, 99.74% European or greater) did not have any genetically Native American ancestors in the past five generations." They consider components below 0.26% to be within the realm of statistical noise. Admixture at levels higher than that (in someone of predominantly European ancestry) is very likely to be real.
Ancestry Painting results will probably be less meaningful for some other populations (e.g., East Africans). Even for West Africans, they've apparently had problems with some African ancestry showing up as "Asian".
I haven't used FTDNA's "Population Finder", but I believe it attempts to break ancestry down into more detailed components (rather than three continental groups). I doubt it can be very accurate at what it's trying to do with the fairly limited reference data currently available. On broader questions I'd expect its accuracy to be pretty comparable to Ancestry Painting's at least for whites. For people whose predominant ancestry isn't European, West African, or East Asian, I could possibly see Population Finder giving better results.
A bit of a disappointing sale at 23andMe today. Details on my blog:
ReplyDeletehttp://www.yourgeneticgenealogist.com/2011/11/23andme-cyber-monday-discount-today.html
What's the margin of error for 23andMe's ancestry painting?
ReplyDeleteI read that FTDNA has a lower margin of error. What do you think?
"What's the margin of error for 23andMe's ancestry painting?"
ReplyDeleteI think it depends on which population you belong to.
For people of Northwestern European ancestry, it seems be very accurate. According to 23andme: "we're able to say with confidence that a person with an all-European Ancestry Painting (actually, 99.74% European or greater) did not have any genetically Native American ancestors in the past five generations." They consider components below 0.26% to be within the realm of statistical noise. Admixture at levels higher than that (in someone of predominantly European ancestry) is very likely to be real.
Ancestry Painting results will probably be less meaningful for some other populations (e.g., East Africans). Even for West Africans, they've apparently had problems with some African ancestry showing up as "Asian".
I haven't used FTDNA's "Population Finder", but I believe it attempts to break ancestry down into more detailed components (rather than three continental groups). I doubt it can be very accurate at what it's trying to do with the fairly limited reference data currently available. On broader questions I'd expect its accuracy to be pretty comparable to Ancestry Painting's at least for whites. For people whose predominant ancestry isn't European, West African, or East Asian, I could possibly see Population Finder giving better results.