Saturday, 23 April 2011

Debunking Mozabites and Company - again

As a first step towards uncovering structure in Fertile Crescent populations, I have run a few unsupervised runs of the region's populations. In these you generally get Mozabite, Saudi or Bedouin, and Druze or Armenian modal components. Egypt generally comes out as a population combining all elements available in largish amounts. This could have two explanations. Either Egypt is where Mozabites, Saudis and Druze all arranged to have a party by the Nile, or Egypt is actually the Mother of them all (there's likely a Father as well) and ADMIXTURE is simply rendering different levels of parental admixture into distinct components. This would happen because Egypt itself has a dominant non-African component mixed with an important African one. Since this component doesn't exist unadmixed anymore, at K=2 with Yoruba and say Basques, you do get an African vs non-African division in the populations, but it's lost at higher Ks, suggesting Yoruba or Masai are not good proxies for it.
In a previous run, I found evidence of a native NorthEast African element in the region.

So I've decided to study North Africa alone for a start. I'll study the Saudi and Druze dominant components later.

Firstly unsupervised run of North Africans with Masai, Yoruba and Basques. I've removed a few problematic Masai and Tunisian individuals, who form their own component:

Notice how regions with especially fertile soils (Tunisia and North Morocco) have little "West African" whereas more desert areas have larger such components. Many such segments are confined to a few individuals, indicating admixture possibly with desert tribes such as Tuareg. I believe such West African segments are actually aboriginal western Green Sahara foragers and semiforagers, largely replaced by the Nile Wave.
Notice also how even in Morocco you get a lot of East African, just like in Egypt, even though southern populations are exclusively West African.

Genetic Distance estimates (Fst):
Nile 1st Wave to East African 84
Nile 1st Wave to Basque 51
Nile 1st Wave to West African 124
Basque to East African 127
Basque to West African 177
East African to West African 44

So what's happening here? The Nile component, higher in Tunisians and many Moroccans than in Egyptians themselves, has actually more affinity to East African than to geographically closer West Africans. It also may have a lot of Basque-like (Mesopotamian Core) admixture.

My point is: Mozabites and other North Africans plot in MDS distantly to Subsaharan Africans but much closer than their low affinity to West Africans would predict, as shown in this run. I think this is due to a Neolithic North Eastern African migration. In other words, for North Africa it's maybe more due to a more recent "Out of Egypt" than due to an ancient "Out of Africa" event.

Next I'll look into North Africans in supervised mode.

No comments:

Post a Comment