I ran supervised admixture on the Yunusbayev et al dataset from the Caucasus using my reference 3 data to see how the Yunusbayev samples looked in my Ref3 admixture component space.
Here's the spreadsheet for Yunusbayev admixture results. You can compare with the reference 3 results.
Here's our bar chart for Yunusbayev results. Remember you can click on the legend or the table headers to sort.
Hey Zack,
Vasishta directed me here. I had a question about the Tajik's results. I was attempting to calculate the ANI or West Eurasian ancestry in Tajiks using the linear regression forumla given to me by Vasishta. I think others have used it as well. Although, we've assumed it was South Asian specific, it gave me some seemingly accurate results?
I calculated around 79% ANI with 5-6% ASI and 13-15% East Eurasian? Is this accurate? It would put them very close to the results of the HGDP Pashtuns from Kurram Valley and the HGDP Kalash.
Also, take a look at these Dodecad results for Tajiks.
https://docs.google.com/spreadsheet/...QblVRR2c#gid=0
They are 14.7% East Eurasian (Siberian + NE Asian) and 13% South Asian. I think the East Eurasian ancestry is completely there as they are surrounded by Turkic speakers and have close proximity to Western China. I also contemplating that some of the South Asian ancestry is ASI as Vasishta mentioned them being about 4% Onge.
Also, just for your reference, I used these estimates made by you for the linear regression forumla.
https://docs.google.com/spreadsheet/ccc?key=0AuW3R0Ys-P4HdDhib1M5OE1wWENNb2haUFFWZzNBMEE#gid=0
I wouldn't use any linear regression formula to try to compute ANI or ASI for non-South Asian groups. These regressions are based on Reich et al's results and while they can be considered somewhat valid in the range of admixture results of the groups Reich et al used, it's not possible to have any idea of the accuracy of these regression estimates outside that range.
The Tajik admixture result are closest to Kalash, Pathan and Burusho among South Asian groups.