Wednesday, 20 April 2011

Unsupervised Northern Eurasia+new samples part 3

In this last part of the unsupervised run I'll present results of New World populations and participants, and some South Asians I introduced into the run without South Asian populations (to avoid the artificial Irula modal component).

Mexico1 and Mexico2 are demonstrative of the high variability still found in Mexicans, a young population in terms of ethnogenesis from mostly Spaniards and Neolithic Amerindians, with a little West African as well. Unlike ancient admixture events in the Old World which are mostly stabilized with little variation between components in different individuals. Same can be said of Dominican Republic samples (Hispaniola1 and 2) even though here the West African component is larger and the Amerindian smaller. EUDR has major ancestry directly from Europe. You can also see likely continuing admixture in populations with exclusively aboriginal self-identification.

GeUSA has half German ancestry. Some of the USAs are mostly colonial but colonialUSA1 will remain the only one with such a tag since its hard to find Americans without either some colonial or non-colonial ancestry. USA6 is African-American, another very young New World population drawing from diverse Neolithic Core Areas, still with very different admixture levels between individuals.
Afrikaner1 is a White South African sample of Afrikaner ancestry. The Cape has a Mediterranean climate little suited for Bantu crops, but ideal for Fertile Crescent ones. This may explain why Dutch farmers managed to settle and flourish in the forager/pastoralist (San and Khoikhoi)) inhabited area unlike Europeans in other regions of Africa, who had few food producing advantages and local disease resistance over the West African Neolithics.

As for the South Asian samples. I welcome samples from the area, was a little surprised to get 3 Tamil Brahmins. SIndia1 is mostly Tamil from Sri Lanka. UKIND is British with some Indian ancestry.
Interestingly the Fertile Crescent element in the Tamil samples isn't very different here.
The remaining components I tend to believe are artificial. Ancient South Indians (ASI) were likely a Coastal Migration derived population. Yet Papuans are separated by seas, thousands of kilometres and tens of thousands of years and the genetic affinity may not be close enough, so this ASI component is IMO being split artificially between the remaining poles. This shouldn't be considered evidence of admixture with the pole modal populations though. It possibly simply means that ASI is somewhat distantly intermediate between them. I'll look closer to South Asian populations later.
But before I want to search for finer structure in the Fertile Crescent Cores.

Here's the spreadsheet. Don't take small components literally, especially in an unsupervised run. They may be real, but also noise or may be replacing something different not represented in the other 6 components.


  1. Can you please mention the reference populations/K's against the colored squares? I know that some components are likely noise, but it'd be nice to know them anyway. It's sort of tedious trying to find out where each component is modal via the spreadsheet.

  2. I'm not naming them since ADMIXTURE picked them up not me. This is unsupervised. You can check previous posts to see which population has the largest amount.

  3. All 3 posts are from the very same run.