search this blog

Tuesday, June 12, 2018

Dali_EBA and West_Siberia_N in qpGraph


Below is a qpGraph tree that I've been working on for a while. I'll be posting more output from random analyses like this from now on. The relevant graph file is available in the zip folder here. Any ideas what else can be done with this topology?

See also...

Graeco-Aryan parallels

30 comments:

Lenny Dykstra said...

Is it possible to fit Goyet or WHG into this topology?

The "Eurasian 4/Eurasian 5" ghost population is intriguing-- a cousin of ANE that is responsible for the non-Basal portion of CHG's (and presumably Iran_N's) ancestry. But also interestingly this population is attributed 2/3 of EHG's ancestry with ANE only 1/3. This might be why Khvalynsk's CHG-related ancestry was initially overlooked-- CHG and EHG both share considerable ancestry from this same Paleolithic ghost pop.

Two possible explanations for this topology:

Scenario #1: Eurasian 4/Eurasian 5 were spatially located between Goyet-dominated Western Europe (35kya) and Ust-Ishim types in Siberia (45kya). IOW, they occupied the steppe/forest zone of Eastern Europe in the Paleolithic.

Eurasian 4 split off (30kya?) and migrated down thru the Caucasus, running into Basal Eurasians and thus giving rise to CHG/Iran_N. Their northern "Eurasian 5" cousins who remained on the forest/steppe zone absorbed an "Eastern" pulse from ANE mammoth hunters in the Mesolithic (~30-15kya?), giving rise to EHG.

Scenario #2: Eurasian 4/5 split up while heading NORTH from the Levant/Anatolia, with Eurasian 4 remaining in the Near East/Caucasus to blend with BE and Eurasian 5 continuing to the E Euro steppe/forest zone north of the Caucasus. This scenario would require Basal Eurasians as a "late" arrival in Levant/Fertile Crescent/Iran (post-20kya), since EHG didn't harbor any BE ancestry. Which implies BE was bottled up in the Arabian Peninsula, North Africa, and/or East Africa thru most of the Paleolithic.

Shaikorth said...

@Lenny Dykstra
CHG has looked a bit WHG-shifted compared to Iran_N in some previous tests. The survival of Ust-Ishim level or more basal (but not "basal eurasian") populations in Kostenki/MA-1 times and significantly contributing to proto-WHG's may be unlikely though, Reich/Lipson just needed some ENA to fit WHG and ANE in their tree with Kostenki. EHG fits as Eurasian5 + MA-1 related which in turn needs some Ust-Ishim level mixture here but it also fits fine as WHG+West_Siberia_N in qpAdm (better than as WHG+AG3) and that might resolve the need for more basal contributions. The only examples of Goyet's contemporaries in European Russia were the Sungir folks, who were Kostenki14-like.

Lee Albee said...

@davidiski

What is your goal? What hypothesis are you wanting to test with this? That would dictate what sort of suggestions I would give here.

Are you trying to demonstrate that western East asia has Western Eurasian genetics?

It would be interesting if Afanasievo or Adronova people fed into the Dali_EBA?

based on Early bronze age I would lean toward Afanasievo would be closer than Adronova.

Though modern Bai people would probably have more diverse ancestry as waves of Steppe people went back and forth

Lee




Davidski said...

@Lee Albee

It would be interesting if Afanasievo or Adronova people fed into the Dali_EBA?

As far as I can tell, it's unlikely, because Afanasievo is at least 50% EHG, and I wasn't able to model Dali_EBA with EHG mixture using this topology.

Samuel Andrews said...

The bulk of central Asia has received lots of new migrants since the time of Dali EBA. But it looks like the area around and near Lake Baikal has a decent amount of continuation in the last 8,000 years. There are lots of modern pops there of mostly Lokomotive-like decent.

Davidski said...

@Lenny Dykstra

Is it possible to fit Goyet or WHG into this topology?

Yes, but only WHG, simply by copying the topology leading to Iron_Gates_HG, but this does worsen the Z score and bumps up the East_Eurasian admixture in Dali_EBA and West_Siberia_N to rather unrealistic levels, which suggests that there's a better way of doing it.

https://1.bp.blogspot.com/-hq4GdEzZLy0/WyDrimpt2JI/AAAAAAAAG5I/FWatJajYmowJliW-_Ln0M7NOG4sHgS7IwCLcBGAs/s693/WHG_qpGraph.png

Not sure what that might be though. Feel free to have a go and I'll run it for you. Goyet might need some archaic ancestry to work.

Salden said...

Them's fighting words:

>Aryan Invasion Theory: Harappan site of Rakhigarhi: DNA study finds no Central Asian trace, junks Aryan invasion theory

https://economictimes.indiatimes.com/news/politics-and-nation/harappan-site-of-rakhigarhi-dna-study-finds-no-central-asian-trace-junks-aryan-invasion-theory/articleshow/64565413.cms

Davidski said...

Aryan Invasion Theory: Harappan site of Rakhigarhi: DNA study finds no Central Asian trace, junks Aryan invasion theory

Well, obviously, that's what was always predicted within the framework of the AIT, so the AIT is doing fine. Can't say the same about OIT though.

velvetgunther said...

@Salden
What a bizarre article drawing bizarre conclusions.

Lee Albee said...

@davidski

Which paper did the Dali_EBA samples come from for this analysis? I want to look at the archeology on them to get some ideas

FrankN said...

What happens to both the "original" and "w. WHG" scenarios if Han are replaced by a more NE Population (eventually even Kennewick)?

Davidski said...

@FrankN

I can't replace the Han with Siberian and especially not Amerindian samples because they have significant levels of ANE ancestry compared to the Han, so that needs to be accounted for. However, I can add them to the tree as mixtures of the ANE and East Eurasian nodes, like Kennewick here. It's a decent fit, but the tree does look a bit wobbly, suggesting that there's a better way of doing this and/or Kennewick is too low quality to produce robust results in such a complex topology.

https://4.bp.blogspot.com/-zmj5hncpr6s/WyHByURKhkI/AAAAAAAAG5Y/15hj_pvxzzMnqgiLkDmprkcPILkMztOkgCLcBGAs/s698/Kennewick_qpGraph.png

@Lee Albee

Dali_EBA is represented by one individual from the South and Central Asian paper. See here...

http://eurogenes.blogspot.com/2018/03/andronovo-pastoralists-brought-steppe.html

https://reich.hms.harvard.edu/datasets

Garvan said...

Blogger Lee Albee said...

Which paper did the Dali_EBA samples come from for this analysis? I want to look at the archeology on them to get some ideas

You will find more information on the sample in "The Genomic Formation of South and Central Asia" online supplementary materials.

https://www.biorxiv.org/content/biorxiv/suppl/2018/03/31/292581.DC1/292581-1.pdf

Around line 700


Balaji said...

@Davidski,

I think what Rai was telling the reporter was that there was considerable ANE detected in 2600 BC. Rakhigarhi (about the same as in present day inhabitants of the region). But this could not have come from the Steppe since Narasimhan et al. write, “Importantly, in the 3rd millenium BCE we do not find any individuals with ancestry derived from Yamnaya-related Steppe pastoralists in Turan. Thus, Steppe_EMBA ancestry was not yet widespread across the region.”

Lee Albee said...

@Davidiski

You may get better results if you use Anzick than Kennewick. But maybe not.

Davidski said...

@Balaji

South Asians have Steppe_MLBA ancestry. You know, Sintashta, Andronovo, etc.

And this is what's missing in the Rakhigarhi samples, as well as in Iran/Turan before the 3rd millenium BCE.

Other types of ANE-related ancestry were already in the region, but so what, because they have nothing to do with R1a-Z93 and Indo-Europeans.


Shaikorth said...

Could someone run these D-stats, mostly to test certain datings:
Mbuti Iran_Hotu MA-1 West_Siberia_N
Mbuti Iran_Hotu EHG West_Siberia_N
Mbuti WHG MA-1 West_Siberia_N
Mbuti WHG MA-1 Botai
Mbuti Dali_EBA West_Siberia_N Botai
Mbuti EHG MA-1 West_Siberia_N
Mbuti EHG MA-1 Botai

Mike the Jedi said...

@Dave

In case you haven't heard yet, genomes from the Fregel North Africa paper are now available...

https://www.ebi.ac.uk/ena/data/view/PRJEB22699

Lenny Dykstra said...

Shaikorth: "The only examples of Goyet's contemporaries in European Russia were the Sungir folks, who were Kostenki14-like."

Ahh, I had forgotten about Sunghir/Kostenki... that makes Scenario #1 unlikely since it didn't seem like Kostenki/Sunghir's descendants made a big impact on moderns (minimal continuity between Sunghir and WHG/EHG)

That makes Scenario #2 more likely: IOW, Eurasian 5 swept into Balkans/E Europe from the south; Kostenki/Sunghir types were replaced sometime in the late Paleolithic. Meanwhile their cousins (Eurasian 4) remained in the MidEast and admixed w/Basal Eurasians incoming from the Arabian Penninsula/East Africa/North Africa.

Frankly, Scenario #2 also explains the following 2 facts better as well:

- WHG showing a "Middle Eastern shift" relative to Goyet (since Eurasian 5 is basal to WHG, and E5 shared ancestry w/Eurasian 4 which admixed with BEs

- The theory of Para-Eurasian/early Basal Eurasian ancestry being present in SSAs is easier explained if BEs/Para-Es spent much of the Paleolithic in N/E Africa.

Shaikorth said...

Sunghir/Kostenki were well separated by time from Mesolithic populations like the WHG's, but it actually is possible that they or something related ended up contributing to MA-1-types and WHG (and EHG by proxy)
See:
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5400393/figure/msw293-F1/
Reich&Lipson mention this kind of model works for WHG too.

Balaji said...

@Davidski

But if the ancient Rakhigarhi people were practically identical to the modern Rakhigarhi people, that would be evidence AIT is false.

I have posted several comments including the following in Bioarxiv for the Narasimhan et al. paper.

“You write that people from the Steppe migrated to Turan, presumably via the Inner Asian Mountain Corridor and mixed to a limited extent with the people there from 2300 – 1500 BCE. The evidence of mixing that you have presented are some outlier individuals from Turan who have apparent Steppe admixture and some outlier individuals from Steppe burials who have Turan admixture. However you also write that the Steppe people hardly mixed with the BMAC people but instead mixed in a major way with people in the Indian Subcontinent. This is very curious because the Indus Valley then as now supported a much larger population than Turan or the Steppe. So quite a large number of Steppe people must have moved through Turan to the Indus Valley to have effected this change. During this migration would they not have had a very significant and detectable effect on the much smaller populations of the Turan on the way?.”

The idea here is that if hordes of Aryan barbarians moved south from the Steppe to invade India, they would have gone through Turan. They would have had to live off the land and the granaries and livestock of BMAC would have been irresistible. It is almost inconceivable for barbarians not to pillage and rape, the effect of the former would show up in the archaeological record and the latter in the aDNA. The fact that BMAC was left unmolested shows that there was no AIT.

Matt said...

@Mike, just looking from the perspective of the Fst scores in the supplement:

PCoA: https://imgur.com/VjSpCb9 . The first dimensions show that IAM and Taforalt look drifted away from Yoruba+Mbuti in a way mostly orthogonal to Eurasians, though partially correlated with Western farmers

Direct comparison of Fst distance between IAM vs Taforalt (the Iberomaurasians): https://imgur.com/a/zAPBjJg . It looks like IAM may be closer to Eurasians than Taforalt, while Taforalt may be closer to Natufians. I am a bit suspicious of the outlying position of TOR's scores here since in non-Fst analysis they are simply Iberia_MNChl alikes.

Guanche vs Mozabite (the closest neighbour joining modern North African pop): https://imgur.com/N6nKQlW. The Guanche population in general looks more isolated and drifted, however, *relatively* closer to IAM / Taforalt (relative to Yoruba+Mbuti / Eurasians).

Distance of Europe_MNChl against Steppe_EMBA: https://i.imgur.com/X0AsQln.png. Equidistant to IAM, Mbuti and Yoruba (very slightly non-equidistant to Taforalt). Most other populations less equidistant.

Han vs WHG: https://imgur.com/91gbQfW . IAM and Taforalt equidistant to WHG and Han, but slightly closer to WHG than Han, relative to isocline based on Yoruba and Mbuti.

Comparison of the IAM and Taforalt may be able to tell us something about early Near East history. They seem reasonably diverged based on Fst.

Salden said...

https://onlinelibrary.wiley.com/doi/pdf/10.1002/ajpa.23607

>New genetic evidence of affinities and discontinuities between bronze age Siberian populations

>Clémence Hollard, Vincent Zvénigorosky, Alexey Kovalev, Yurii Kiryushin, Alexey Tishkin, Igor Lazaretov, Eric Crubézy, Bertrand Ludes, Christine Keyser

>Abstract

>Objectives

This work focuses on the populations of South Siberia during the Eneolithic and Bronze Age and specifically on the contribution of uniparental lineage and phenotypical data to the question of the genetic affinities and discontinuities between western and eastern populations.

>We performed molecular analyses on the remains of 28 ancient humans (10 Afanasievo (3600–2500 BC) and 18 Okunevo (2500–1800 BC) individuals). For each sample, two uniparentally inherited systems (mitochondrial DNA and Y‐chromosome DNA) were studied, in order to trace back maternal and paternal lineages. Phenotype‐informative SNPs (Single Nucleotide Polymorphisms) were also analyzed, along with autosomal STRs (Short Tandem Repeats).


>Most of the Afanasievo men submitted to analysis belonged to a single sub‐haplogroup, R1b1a1a, which reveals the predominance of this haplogroup in these early Bronze Age populations. Conversely, Okunevo individuals carried more diverse paternal lineages that mostly belonged to Asian/Siberian haplogroups. These differences are also apparent, although less strongly, in mitochondrial lineage composition and phenotype marker variant frequencies.

Lee Albee said...

@ Lenny Dykstra

When looking for the source of BE it is going to be pretty complicated

It is fairly certain that they expanded out of their homeland during the last ice age (near the end)

when looking at in that time frame East africa, Southern Arabia peninsula and Southern Iran up to the mountains were fairly all one continuous landscape. Red sea was fairly non-existent as was the Persian gulf. People were probably going back and forth over that range for a long time. Isolated from populations north of the Arabian Desert zone, so perhaps isolated from mixing with neanderthals.

Lee Albee said...

@ Matt
Taforalt ancestry is skewed by SSA ancestry present. If you want to have a better look at what the BE Source population of Taforalt looked like and was really related to, try masking out African ancestry.

Davidski said...

@Balaji

But if the ancient Rakhigarhi people were practically identical to the modern Rakhigarhi people, that would be evidence AIT is false.

They weren't practically identical. They lacked Sintashta-related ancestry, including R1a-Z93, which is very important today in the Indo-European-speaking populations of India, and especially in Indian upper castes.

Why are you still debating these things? Do you really think that you'll change my mind by arguing the opposite to what all of the evidence shows?

Balaji said...

@Davidski

You and the Reich lab folk expect that the ancient Rakhigarhi people will differ from modern people living there in lacking Steppe_MLBA ancestry. Soon the evidence will be published. No fancy analysis will be needed. A PCA with moderns and ancients will be sufficient. Prof. Shinde, a highly respected archaeologist who has capped his career with the excavation of Rakhigarhi, must be privy to the results which is why he has so confidently asserted to the reporter that here was no Aryan invasion.

One place Narasimhan et al. went wrong was in assuming that the Indus_Periphery samples were typical of the Indus Valley. These people were indeed emigrants to Turan from India but they originated in the southeastern frontier of the Indus Civilization and were not from the mainstream. Sarazm_EN or even Gonur1_BA_o, both of which have much higher ANE, might be better proxies for modeling the Indus_Diaspora. Another mistake they made was in assuming that the Swat Valley people were in any way representative of people of the Subcontinent. You have shown that they are outliers with anomalously high levels of Anatolian agriculturalist ancestry.

http://eurogenes.blogspot.com/2018/05/the-protohistoric-swat-valley-indo.html

Matt said...

@Lee Albee; how would you propose to do this?

Davidski said...

@Balaji

Sarazm_EN and Gonur1_BA_o can't explain the Steppe_MLBA ancestry in South Asians, because they have significant West_Siberia_N-like ancestry, which is very different.



aniasi said...

@Davidski

Your notes on the Swat Valley sample has me thinking about something... what if the Indo-Aryans didn't all arrive from the Northwest, but instead with some waves through the Pamirs and Kashmir?

With the Inner Asian Mountain corridor being the best migration route thus far, might we countenance the IA arrival happening through Kashmir and what is now Himachal Pradesh?