The ADMIXTURE results for the Harappa participants (up to HRP0180) for the Reference 3 + Yunusbayev dataset are in a spreadsheet and can also be seen in the bar charts below.
Do take a look at K=12 and K=17 (lowest crossvalidation errors) as well as K=15.
Hi Zack, thanks for this set of runs. The North Indian component at K=15 is very interesting, as it is centered in Baluchistan. In that regard, it is very similar to the Central-South Asian component inferred inferred in Dienekes' ADMIXTURE analysis of Eurasian populations with K=15. My only concern is that the Kalash's lack of outbreeding is distorting some of the results. At K=12 the labeled West-Asian component seems to be modal in them. This really looks like a yet-to-be broken down "Kalash" component. Could you perhaps run K=12, 15 and 17 without the Kalash?