Peeling the population genetic Indian onion

There's a new paper in The American Journal of Human Genetics, Shared and Unique Components of Human Population Structure and Genome-Wide Signals of Positive Selection in South Asia. It's free, so go read it. I don't have time to comment in detail, but I did read the paper, and I want to mention a few things: 1) If you follow Harappa Ancestry Project or Dodecad Ancestry Project the ADMIXTURE and PCA won't be surprising. They'll be familiar. Though the researchers got some nice additional populations in Uttar Pradesh it didn't change the general outlines of what you can already ascertain with the public data sets. 2) The authors seem to de facto ignore the argument in Reconstructing Indian population history that ADMIXTURE components can themselves be decomposed into further real elements. (they acknowledge it in the text, but it doesn't go any further) This is obvious as you move up the K's, and a given component collapses into two obvious elements. More precisely, it may be that the modal South Asian component in their runs is actually a composite, as may be the the minor component. I suspect this is the case, because very low caste tribal elements in South India which show up as "pure" South Asian in some of these ADMIXTURE runs nevertheless may carry West Eurasian markers, such as the derived variant of SLC24A5. 3) The big finding the paper is that the West Eurasian regions of the genome in South Asians may be more diverse in terms of haplotypes than in Europeans, and some extent Near Easterners. The inference here then is that perhaps "Ancestral North Indians" are a source population for other West Eurasians! I think more likely this is not the case. The authors note that it could be due to large effective population size in South Asia. But I think it may have something to do with the fact that the regions of the genome they selected are more admixed and composite than the authors assume. In other words, the diversity is elevated by the bleed-in of "Ancestral South Indians" into this genetic background. There's some interesting stuff about natural selection and diabetes in the paper, but I'll leave that for later.