Originally Posted by
Joqool
I think its mainly Ethiopians who are 50/50 while Somalians are more like 60% SSA and 40% West Eurasian. While Eritreans seem to be closer to 60% West Eurasian and 40% SSA. There was even one Eritrean sample in G25 who was closer to 65% West Eurasian.
In the G25 model below, Somalis and Oromo and Wolayta are all modeled as between 50 and 60 percent Dinka. They got around 5 percentage points lower SSA ancestry when I used Bantu_Kenya instead of Dinka as the SSA source.
In a global ADMIXTURE run with 3 components, Somalis got 55% of the SSA component. Only Capoid-Bambutids and South African Bantu mixed with Khoisan got 100% of the SSA component, but if I wouldn't have included them in the run, it would have slightly increased the amount of SSA ancestry in mainstream SSAs. Hadza got 4% of the Mongoloid component, which maybe means that they have undifferentiated Eurasian ancestry, or that they have ancestry related to the East African branch that was ancestral to Eurasians, or that they share a part of their non-Capoid-non-Bambutid ancestry with Mongoloids. Even some Bantu got 1% of the Mongoloid component in this run, which might be because of the latter reason. This shows all populations from the run with 10% or more of the SSA component:
100 0 0 Mbuti
100 0 0 Khomani_San
100 0 0 Ju_hoan_North
100 0 0 Biaka
100 0 0 BantuSA
99 1 1 Namibia_Bantu_Herero
99 1 1 Malawi_Chewa
99 0 1 Esan
99 0 0 YRI
99 0 0 MSL
99 0 0 ESN
98 1 1 Yoruba
98 1 1 Mende
98 1 1 Malawi_Yao
98 1 1 Malawi_Tumbuka
98 1 1 Malawi_Ngoni
98 0 2 BantuSA_Ovambo
97 2 0 GWD
96 3 1 Mandenka
95 5 1 Luo
95 4 1 BantuKenya
94 5 1 LWK
93 6 1 Luhya
92 8 0 Gambian
91 9 0 Khomani
90 8 2 Dinka
90 10 0 ACB
84 12 4 Hadza1
80 19 1 Kikuyu
78 21 1 ASW
78 20 1 AA
72 27 2 Masai
66 32 2 Datog
55 44 1 Somali
45 55 1 Jew_Ethiopian
45 54 1 Afar
41 59 0 Eritrea
29 68 3 Yemeni_Desert2
24 76 0 Saharawi
24 76 0 Moroccan
23 77 0 Mozabite
23 77 0 Burbur
22 78 0 Algerian
21 79 0 Tunisian
21 78 1 Yemeni_Highlands_Raymah
19 81 0 Shaigi
18 81 0 Libyan
17 83 0 Ain_Touta
17 81 2 Yemeni
17 19 64 Papuan
14 86 0 Egyptian
14 74 12 PUR
13 23 64 Australian
11 34 56 PEL_SomeAfrican
10 90 0 BedouinA
10 89 0 Yemeni_Highlands
10 18 71 Nasioi
Originally Posted by
Joqool
Could it be that the Hadza in 1240K+HO set are drifted? It's peculiar because in G25, Hadza seem to show minor Eurasian (Caucasoid) admixture:
I think Hadza just represent one of the major human races. Even humanphenotypes.net says that the Hadza type is "Very distinct from any other type in the world and has probably long been isolated."
In a paper from 2018 titled "Genetic Ancestry of Hadza and Sandawe Peoples Reveals Ancient Population Structure in Africa", Hadza had an FST distance of around .1 or more to every other African population group (https://academic.oup.com/gbe/article/10/3/875/4935243, Table 2):
FST distance to Hadza:
.096 Nilo-Saharan
.099 Eastern and Southern Bantu
.099 Western Niger-Congo
.101 Omotic
.108 Western Pygmy
.110 Sandawe
.114 Berber
.122 Eastern Pygmy
.122 Khoisan
.125 Cushitic
.129 Arabian
The same paper estimated that Hadza split off from Niger-Congo and Nilo-Safaran Africans before Eurasians did:
We estimated divergence times of ∼98,000 to ∼96,000 years for Hadza ancestry from Eastern Pygmy and Khoisan ancestries, respectively, followed by divergence times of ∼89,000 and ∼88,000 years for Western Pygmy and Sandawe ancestries, respectively, and then followed by divergence times of ∼81,000 to 76,000 years for Arabian, Berber, eastern and southern Bantu-speaking, Nilo-Saharan, and Western Niger-Congo ancestries (table 3). These divergence times are all before Out-of-Africa, and therefore support early divergence of Hadza ancestry. In contrast, we estimated divergence times for Sandawe ancestry of ∼55,000–34,000 years (table 3).
Also Hadza have more Y-DNA B and less E1b1, but Sandawe have more E1b1 and less B:
In the Hadza population, the distribution of Y chromosomes includes mostly B2 haplogroups, with a smaller number of E1b1a haplogroups, which are common in Niger-Congo-speaking populations, and E1b1b haplogroups, which are common in Cushitic populations (Tishkoff et al. 2007). In the Sandawe population, E1b1a and E1b1b haplogroups are more common, with lower frequencies of B2 and A3b2 haplogroups (Tishkoff et al. 2007). Using autosomal data, Tishkoff et al. (2009) concluded that the Hadza population had ∼72% ancestry distantly related to Khoisan and Pygmy ancestries, with ∼22% Niger-Congo ancestry and ∼6% Cushitic ancestry.
Originally Posted by
Joqool
You should also add Dinka or Sudanese to the run. They are predominantly Nilotic with some West-Central African ancestry but no West Eurasian. Meanwhile Luo are so heavily mixed with Bantus that they are now mostly West-Central African with significant Nilotic admixture.
Oh yeah, you're right, Luo even look like Bantu. On G25, they're slightly closer to Yoruba than to Dinka or Mursi.
1240K+HO has one Dinka sample, and there's some samples of Shaigiya (who are Arabized Nubians from Northern Sudan), but otherwise it's missing samples from Sudan. There's also no samples from the SNNPR of Ethiopia (Southern Nations, Nationalities, and People's Region), even though G25 includes Mursi from the SNNPR.
I guess I'll have to start merging datasets instead of just relying on 1240K+HO.
Bookmarks