Monday, February 12, 2018

More on Admixture Testing

Don’t compare apples to oranges.

I will attempt to explain the problems about parental populations and genetic tests for the typical Type I “movement” Nutzi dimwit. I will use simplified examples to illustrate the concept.

Let’s say Company X is assaying the autosomal genome of a Russian and trying to construct “ancestry percentages.”  The parent populations they are using as the standards of comparison are Germans and Japanese.  The Russian tests out as “95% German and 5% Japanese” (cue heavy breathing from the anti-Slavic contingent of Der Movement, Inc.).  However, if that same Russian was tested with parental populations of Russians and Japanese, the test results may be “100% Russian and 0% Japanese.”  The same basic principle applies to other groups.  The more similar the person or group tested is to the parental populations, the less “exotic admixture” they will display, and vice versa.

This does NOT mean that “race is a social construct” or “genetics is subjective” or “deconstructivism is correct.”  In the example above, the Russian’s genome is what it is, and can be identified as ethnically Russian.  However how one wishes to describe the objective reality of the genome can be subjective, or partially subjective, dependent on what parental populations are used.  And, even more fundamentally, how those parental populations are named.  What if the German population was labeled as “European” and the Japanese as “Asian.”  Then the Russian is “95% European and 5% Asian.”  On the other hand, most Germans would test as “100% European” as they are essentially being compared to themselves (in practice  of course there is statistical error, and not all Germans derive from identical genepools, so typical German percentages may vary from, say, “98-100 European” in this case.  The main point is that if Germans are being tested against a German parental population that is labelled as “European” they will have higher European percentages than other groups).

Of course, one can argue that this critique can be taken too far.  If you want to know the admixture percentages of Puerto Ricans, then using, say, Iberian, West African, and Amerindian parental populations are reasonable.  Using Puerto Ricans themselves as the parental population, and comparing Puerto Ricans to Puerto Ricans – with people getting test results of “98-100% Puerto Rican” is not going to be informative about the admixture question asked.

Fair enough.  But what if we were asking: which Hispanic group has the most admixture?  And then assume you use as the parental populations Puerto Ricans and Japanese.  Mexicans are going to show more “Asian admixture” in this case, given their greater percentage of Amerindian ancestry compared to Puerto Ricans.  If the parental populations were Mexicans and Nigerians, then Puerto Ricans would be seen as a Mexican-Nigerian mix.  

You can see that the manner in which the question is asked, and what data are used to answer it, is going to influence the interpretation of the outcome.

Subtleties like this fly right over the head of the typical “movement” fetishist.