September 20, 2005

Blind test of DNA BioScience

DNA Bioscience is a British firm which offers (among other things) the Ancestry By DNA test. Recently, the company was featured in a Guardian story, in which four journalists from the paper submitted their DNA samples and the company predicted their skin color and ethnic origin. Person A was an Eastern European Jew who was predicted as being European with East Asian admixture, a reasonable finding, since both Ashkenazi Jews and eastern Slavs have some degree of Asian admixture. Person B was a Sub-Saharan African who was classified as such, again a reasonable result. Person C was a native Briton who was however given minor Native American ancestry, again underscoring the tendency of the test to wrongly predict such ancestry in persons without any possible Native American connection. Person D was a native Indian. The company shrewdly inferred his origin, even though a European-East Asian-Sub-Saharan-Native American mix does not reflect the origin of Indians. The suggestions that the individual had East Asian grandparents or his ancestors reached Britain through Africa (to explain the Sub-Saharan component) are of course rationalizations, and a good example of a case where this type of test is not useful.

All in all, this small test establishes once again the strength of admixture testing in predicting majority ancestry, its overreporting of minority ancestry that does not exist, and its difficulty in dealing with populations that fall outside the group of a priori chosen populations.

No comments: