August 28, 2004


AncestryByDNA has released a EURO-DNA test which reports percentages of "Northern European," "Southeastern European," "Middle-Eastern," and "South Asian" admixture based on a 320 ancestry-informative markers (AIMs).

The ad-hoc choice of the four ancestral groups and the rather confusing commentary and/or anomalous results (Iberians on average ~16% "South Asian"?) may discourage many from taking the test, especially at a price tag of $399. Still, EURO-DNA is a step towards personalized genetic archaeology, even though the theoretical assumptions and methodology leave much to be desired at this stage.


If you start with the a priori breakdown into 4 groups, then each individual will have 4 numbers that add up to 100%. One could just as easily have used a "Southwestern European", "Northeastern European", "Middle Eastern" and "South Asian" breakdown, and again each individual would have 4 numbers adding up to 100%.

The trick is to start with a collection of individuals, remove identifying tags and cluster them, thus identifying the real genetic components in the population, if any such components can be detected. This was the procedure followed by Rosenberg et al. [1]. In that analysis, wholly different clusterings emerged, with e.g., the specificity of Iberian Basques, who were allocated their own cluster, was discovered.

By contrast, an Iberian Basque taking the EURO-DNA test would perhaps get a score high in NOR/MED which however obfuscates the real genetic structure of the Basque population which is highly specific, as the Basques are an ancient ethnolinguistic isolate of the Iberian peninsula rather than the product of "admixture".

AncestryByDNA must show why its chosen four-group breakdown is used in lieu of other potential choices.

[1] Rosenberg et al (2002)

Update #2: Check out the comments for some additional information by Dr. Tony Frudakis of DNAPrint who is involved in the creation of EURO-DNA 1.0 and the AncestryByDNA tests.

No comments: