In a previous post I summarized extensive evidence by myself and Turkish researchers to the effect that modern Turks are about 1/7 descended from Central Asian Turkic speakers, and 6/7 from pre-Turkic West Asians.
Some people have argued that Uzbeks, the best representative of the Central Asian ancestors of the Turks are inappropriate as a parental population.
Can Turks be modeled as a 1/7-6/7 simple mix of West Eurasians and Central Asians? I refer to my most recent K=11 ADMIXTURE results as useful data that can be used to test this hypothesis once again.
I will use the 4-way average of Greek_D, Armenian_D, Georgians, and Syrians as representative of the "West Eurasian" component in Turks. These 4 populations border Turks from the West, East, North, and South, and their average is expected to be a good stand-in for what pre-Turkish Anatolians were like, and probably more robust than choosing arbitrarily just one of the 4 populations.
I will use Uzbeks as representative of Central Asian Turks, and I will calculate the weighted average of the two (1/7 Uzbek + 6/7 "West Eurasian"). I will then compare this with the average of the Turks (from Behar et al. 2010)+Turkish_D combined sample.
If Turks can be modeled as the simple mix I have claimed, then the empirical Turkish average will be similar to the simulated one (1/7 Uzbek + 6/7 "West Eurasian"). Here are the actual numbers:
As you can see, the simulated average is virtually identical to the empirical one. All components do not deviate from it by more than 0.4%, and only the most important West Asian one deviates by a mere 1.8% which, in relative terms (divided by the mean of 49.2%) represents a 3.7% error.
Given the finite sample sizes, the limitations of ADMIXTURE, and the use of a 4-way average as a proxy for pre-Turkish Anatolians, I can easily claim that this does not only confirm the validity of my model but to an extraordinary degree.
A different way of testing the model's validity is the correlation between the empirical and simulated admixture proportions which is 0.99956. I don't think I need to point out how remarkable this is.
The empirical data are consistent with the idea that Anatolian Turks are a simple mix of a West Eurasian population element equivalent to the average of their immediate neighbors, and a Central Asian population element similar to Uzbeks in a 6:1 analogy. These results confirm and extend the extensive evidence of the previous post.
UPDATE (May 21): In a new experiment, I demonstrate that all available Turkic samples fall almost perfectly on a cline between West and East Eurasians. That experiment also shows that Uzbeks are the most West Eurasian out of the available Central Asian Turkic populations.
It is still unclear what the ratio of West/East Eurasian elements in Turkic people who entered Anatolia was, but these results certainly point out that the Uzbeks are not unusually Mongoloid in their makeup among Turkic peoples, rather the opposite.