Looking at South Asia

[Chad Rohlfsen]

I have begun looking at the new samples. Below is a basic tree that I use to build up from and create later samples.

OpeningTree.png

This second graph includes a combination of all three Indus diaspora samples. It is rough, so I will need to go back and clean this up with a few extra nodes, in my opinion. While the paper asserts there is no Anatolian in the Indus samples, looking at ADMIXTURE and qpAdm, qpGraph begs to differ. There is also these gems that led to it... and make it seem likely

worst f-stat: Ana Haj Gan Ind -0.001647 -0.004436 -0.002789 0.000883 -3.158

result: Onge Ganj_Dareh_N Indus_Diaspora -0.009536 0.001749 -5.451 191331
result: Onge Hajji_Firuz_ChL Indus_Diaspora -0.005081 0.001787 -2.844 193475
result: Onge Tepe_Anau_EN Indus_Diaspora -0.009280 0.001798 -5.162 185009
result: Onge Parkhai_EN Indus_Diaspora -0.010915 0.001745 -6.254 190806
result: Onge Geoksiur_EN Indus_Diaspora -0.012715 0.001863 -6.826 160396
result: Onge Sarazm_EN Indus_Diaspora -0.009197 0.001955 -4.705 165486

Indus_Diaspora.png

[Chad Rohlfsen]

In the process of breaking down Central Asia, then seeing if that goes into the individual Indus Diaspora samples.

[Kale]

It's interesting that Anatolian ancestry made it so far East. As I was playing around, similar to your direct F3's, the best simple result I was getting for Indus_Diaspora was like a 75/25 mix of Central Asian neolithic + ENA (though with maybe a bit less Anatolian) The Central Asian neolithics themselves being mostly Iran_N related and with ~20% Anatolian. I think they also have some sort of special relation with CHG (be it actual ancestry or otherwise). Then obviously they have some West_Siberian_N-like mixture.

[ren]

I have begun looking at the new samples. Below is a basic tree that I use to build up from and create later samples.

OpeningTree.png

This second graph includes a combination of all three Indus diaspora samples. It is rough, so I will need to go back and clean this up with a few extra nodes, in my opinion. While the paper asserts there is no Anatolian in the Indus samples, looking at ADMIXTURE and qpAdm, qpGraph begs to differ. There is also these gems that led to it... and make it seem likely

worst f-stat: Ana Haj Gan Ind -0.001647 -0.004436 -0.002789 0.000883 -3.158

result: Onge Ganj_Dareh_N Indus_Diaspora -0.009536 0.001749 -5.451 191331
result: Onge Hajji_Firuz_ChL Indus_Diaspora -0.005081 0.001787 -2.844 193475
result: Onge Tepe_Anau_EN Indus_Diaspora -0.009280 0.001798 -5.162 185009
result: Onge Parkhai_EN Indus_Diaspora -0.010915 0.001745 -6.254 190806
result: Onge Geoksiur_EN Indus_Diaspora -0.012715 0.001863 -6.826 160396
result: Onge Sarazm_EN Indus_Diaspora -0.009197 0.001955 -4.705 165486

Indus_Diaspora.png

What could explain Ust-Ishim's shared clad with ENA, the lack of inclusion of Paleolithic European samples?

[Chad Rohlfsen]

Could be a few things. Lack of a Basal-Eu edge to Western hunters, Ust-Ishim really being early ENA, only equally close to West and East due to excess archaic in ancient Westerners... This was just one solution.

[Chad Rohlfsen]

left pops:
Brahmin
Shahr_I_Sokhta_BA1
Onge
Sintashta
West_Siberia_N

right pops:
Mota
Ust_Ishim
Kostenki14
MA1
GoyetQ116-1
Vestonice16
Villabruna
ElMiron
Iron_Gates
EHG
Ganj_Dareh_N
Natufian
Boncuklu
Steppe_EMBA
Globular_Amphora
Hajji_Firuz_ChL
LBK_EN
Levant_BA

best coefficients: 0.503 0.370 0.021 0.106
std. errors: 0.043 0.035 0.050 0.032
fixed pat wt dof chisq tail prob
0000 0 14 8.163 0.880656 0.503 0.370 0.021 0.106

[Chad Rohlfsen]

left pops:
Brahmin
Shahr_I_Sokhta_BA3
Sintashta
West_Siberia_N

best coefficients: 0.849 0.144 0.007
std. errors: 0.039 0.042 0.031

fixed pat wt dof chisq tail prob
000 0 15 14.652 0.476798 0.849 0.144 0.007
001 1 16 24.724 0.0748407 0.853 0.147 0.000
010 1 16 25.228 0.0659119 0.951 0.000 0.049
100 1 16 834.300 0 0.000 0.850 0.150
011 2 17 31.618 0.0167765 1.000 0.000 0.000

[Chad Rohlfsen]

left pops:
Brahmin
Shahr_I_Sokhta_BA1
West_Siberia_N
Onge

right pops:
Mota
Ust_Ishim
Kostenki14
MA1
GoyetQ116-1
Vestonice16
Villabruna
ElMiron
Iron_Gates
EHG
Ganj_Dareh_N
Natufian
Anatolia_N
Boncuklu
Steppe_EMBA
Globular_Amphora
LBK_EN
Ami


best coefficients: 0.523 0.116 0.361
std. errors: 0.030 0.025 0.029

fixed pat wt dof chisq tail prob
000 0 15 6.921 0.959807 0.523 0.116 0.361

[Chad Rohlfsen]

Malayan looks to best represent the South Indian ancestry of Brahmins. I tried to add some Iranian admixture, but it was -1.5%. So, I may need an even more West Asian source for another attempt. Certainly an interesting output.

left pops:
Brahmin
Malayan
Gonur1_BA
Dali_EBA
Sintashta

right pops:
Mota
Ust_Ishim
Kostenki14
MA1
GoyetQ116-1
Vestonice16
Villabruna
ElMiron
Iron_Gates
EHG
Ganj_Dareh_N
Anatolia_N
Hajji_Firuz_ChL
Ami
West_Siberia_N
Onge


best coefficients: 0.527 0.365 0.062 0.046
std. errors: 0.027 0.040 0.029 0.040

fixed pat wt dof chisq tail prob
0000 0 12 8.702 0.728139 0.527 0.365 0.062 0.046
0001 1 13 11.129 0.600042 0.522 0.397 0.081 0.000
0010 1 13 16.616 0.217443 0.553 0.360 0.000 0.087

[Chad Rohlfsen]

If IVC ends up being just Iran plus something like Onge, with no Anatolian, then watch BMAC numbers jump and Sintashta or Steppe MLBA really decline. We need IVC samples to be sure. But if this Malayan represents IVC, plus a later SA admixture event into Brahmins, then BMAC is going to go way up, like it is here.

left pops:
Brahmin
Malayan
Sappali_Tepe_BA
Krasnoyarsk_MLBA
Dali_EBA

right pops:
Mota
Ust_Ishim
Kostenki14
MA1
GoyetQ116-1
Vestonice16
Villabruna
ElMiron
Iron_Gates
EHG
Ganj_Dareh_N
Anatolia_N
Hajji_Firuz_ChL
Ami
West_Siberia_N
Onge

best coefficients: 0.508 0.364 0.043 0.085
std. errors: 0.028 0.038 0.042 0.031

fixed pat wt dof chisq tail prob
0000 0 12 8.008 0.784502 0.508 0.364 0.043 0.085
0001 1 13 15.695 0.26602 0.548 0.334 0.119 0.000
0010 1 13 9.086 0.766445 0.503 0.392 0.000 0.105