A little while ago I discussed South Asian genetics and where Pashtuns fall on that spectrum. I think the conclusions, while broadly correct, were lacking a bit in detail. In a sense this is inevitable, as there isn’t a great data set I’ve found for Afghan Pashtuns that specifies both tribe and location, while being free of excess Turkic admixture.
That said I decided to do another PCA. This time I have two sets of Pashtun samples. One being Afghan Pashtuns from Kandahar, the other being Pakistani Pashtuns from Kurram. Based on the data I have it appears that location is a stronger predictor of where Pashtuns cluster genetically than tribe, so I went with that grouping. I also removed some non-Pashtun samples that seemed to just clutter the chart, while adding Punjabi Kamboj samples. See here for additional methodology details.