PDA

View Full Version : Caucasian in South Asian



TJRocks760
12-25-2016, 09:14 PM
Hey all,

Could someone help me break down what it means that a lot of those who are identified as South Asian see tons of Caucasian in their admixtures?

I'm reading there is some Kurdish, or Turkic in us. What's the current data on the topic?

I know my Pakstani/Indian history but not really anything on the "other" part of my admixture. I should add that some of the stuff that others talk about goes over my head. If someone could break down to layperson terms.

Arame
12-28-2016, 08:06 AM
Could someone help me break down what it means that a lot of those who are identified as South Asian see tons of Caucasian in their admixtures?


It means You have tons of Indo European ancestry. :P Yamna had some 50% of CHG. While Yamna like component is indispensable to model IE South Asians. Look at Lazaridis et al 2016 for that.

Arame
12-28-2016, 08:08 AM
Here is the model. Yamna related groups are classified as Steppe EMBA in this graph.

http://i63.tinypic.com/11kavwx.jpg

jesus
12-28-2016, 09:23 AM
Majority of it(in South Asia) is related to Iranian Neolithic(and possibly Iran ChL) than to steppe groups

Anabasis
12-28-2016, 01:33 PM
It means You have tons of Indo European ancestry. :P Yamna had some 50% of CHG. While Yamna like component is indispensable to model IE South Asians. Look at Lazaridis et al 2016 for that.

You are very wrong with this assumption. CHG is much more indegious to region of Eastern Turkey- South Caucaus-Iran Zagros line which can be associated with proto -Dravidian speaking pre historic tribes rather then proto Indo European. CHG in Yamnaya might be happened due to interaction with steppe populations and native CHG populations in North Caucaus. On the other hand paternal line of Yamnaya (R1b-Z2103) not so widespread over CHG based modern populations except Assyrians and Armenians. Most probably the reason for this is the Yamnaya Effect on South Caucaus but not CHG effect on Yamnaya.And you might be right on "Yamna like component is indispensable to model IE South Asians".

Gravetto-Danubian
12-28-2016, 01:51 PM
Yes I'd tend to agree. It's better to refer to specifics like Satsurblia-Kotias, Armenian Chalcolithic/ EBA, Iran Neolithic & Chalcolithic.
Some Kotias ancestry came with movements from Armenian Highlands, others with a MBA steppe group. It'll depend on the individual.
If he really wants to know he should submit to Dave's Global plot and we can analyse it.

parasar
12-29-2016, 04:47 PM
Hey all,

Could someone help me break down what it means that a lot of those who are identified as South Asian see tons of Caucasian in their admixtures?

I'm reading there is some Kurdish, or Turkic in us. What's the current data on the topic?

I know my Pakstani/Indian history but not really anything on the "other" part of my admixture. I should add that some of the stuff that others talk about goes over my head. If someone could break down to layperson terms.

IMO this Causcasian/Iran Neolithic is very old and not that related to historical events.

Let's assume that Y-R is related to PIE. Next we have to delineate the two main types of Y-R in South Asia. One Y-R2 is shared with Iran Neolithic and the other Y-R1a is shared with the steppe.

IMO, Afanasevo in conjunction with Rakhigarhi will be the key to understanding how this Y-R movement came about. I have been of the opinion that R coming from the Baikal has been present continuously in the Eurasian steppe region since MA1.

http://www.unzcloud.com/wp-content/uploads/2015/11/dstat.png
http://www.nature.com/articles/ncomms9912

In the above we see the Kotias/CHG like affecting the more southern groups and Afanasevo like affecting the northern ones. It would be a reasonable assumption that the latter is the IE one.

It is also my thinking that this Afanasevo like also moved west and brought R1a to Europe with Corded Ware.
Davidski at Eurogenes at one time had noticed something similar:

One of the most intriguing admixture edges I picked up runs from the branch leading to Afanasievo to the base of the branch leading to Corded Ware (see tree5 below).

If this isn't an artifact of the methods - and I think not, because this admixture edge is also one of the most easily reproducible - then it has some interesting implications. For instance, it suggests that the origin of the Corded Ware Culture was not in the western part of the Yamnaya horizon or nearby, as is often assumed, but much further east, perhaps where the ancestors of the Afanasievo pastoralists began their migration to the Altai region. Maybe somewhere north of the Caspian?...

So this looks like more than just broadly similar admixture events. Also, if it turns out that the Tarim Basin mummies did get their R1a from Afanasievo, then we could be looking at the movements of basically the same people both to the west and east even before Yamnaya got really going.

[To clarify, I believe he does not not support the above scenario.]


We saw some support of a movement from the Inner Eurasian steppe to Europe here: http://biorxiv.org/content/early/2016/12/19/094243
"Y. pestis may have entered Europe from Central Eurasia during an expansion of steppe pastoralists ... Phylogenetic analysis of the two reconstructed Y. pestis genomes from the Altai region shows that they occupy a phylogenetic position ancestral to all medieval and extant Y. pestis strains"

Arame
12-30-2016, 08:18 AM
CHG is much more indegious to region of Eastern Turkey- South Caucaus-Iran Zagros line which can be associated with proto -Dravidian

Anabasis

The IBS map of Kotias

http://s019.radikal.ru/i619/1511/96/6a5d57f4d459.png

Do You see South Indian Dravidians having strong affinity with Kotias? I don't see.

If You don't trust Srkz's map. This is a map from the link posted by parasar

http://www.nature.com/articles/ncomms9912/figures/4

As for the rest. Who said the contrary? Me?


Btw there was no CHG in South Zagros in Neolithic. It came there only at Chalcholithic. The first signs of CHG started to appear at Late Neolithic. Associated with G2a1 Y dna.
Old presence of CHG in South Asia is even more impossible. Only Iran Neo like stuff.

Gravetto-Danubian
12-30-2016, 09:17 AM
delete

Gravetto-Danubian
12-30-2016, 09:18 AM
Arame

The impression I get is that Satsurblia/ Kotias and Iran Neolithic are basically the same population.
It's just that Iran Neolithic has more Onge/Ust-Ishm admixture, and more Basal.
Given that there is a gap between Zarzian levels in Iran and the pre-Ceramic Neolithic ones, a case can be made for a recolonization of the Zagros by a population extremely similar to Satsurblia. This obviously happened already by 10000 BC.

Iran_Neolithic:WC1
"Satsurblia:SATP" 86.85
"Ust_Ishim" 10.45
"Israel_Natufian:I0861" 2.7
"Villabruna:I9030" 0
"MA1:MA1" 0
"AfontovaGora3:I9050.damage" 0

pegasus
12-30-2016, 11:01 PM
Kotias , Iran_N, Iran Hotu are all part of the same family of peoples , with Kotias , CHG having a villabruna shift and Iran_N /Iran Hotu shifting more in SW Asian and South Asian direction.
Ultimately the ethnogenesis of these peoples happened in Iran for sure, local Basal Eurasians mixing up with Malta Buret people arriving from Central Asia.

One interesting I observed through Genetikers phenotype analysis , is Iran_N peoples shared traits more akin to those found in MA-1 , ie dark brown skin, dark eyes and black hair
https://genetiker.wordpress.com/2016/07/18/phenotype-snps-from-ancient-iran/

CHG on the other had the derived SLC24A5 gene and were relatively light skinned.

The so called Caucasus is mainly coming Neolithic Iranians who arrived in South Asia 10 kya and/or the demic diffusion of mixed Steppe and Central Asian/BMAC agriculturalist(also mainly Iran_N descended) populations that diffused across NW South Asia the in the Bronze age.

parasar
12-31-2016, 07:41 PM
Kotias , Iran_N, Iran Hotu are all part of the same family of peoples , with Kotias , CHG having a villabruna shift and Iran_N /Iran Hotu shifting more in SW Asian and South Asian direction.
Ultimately the ethnogenesis of these peoples happened in Iran for sure, local Basal Eurasians mixing up with Malta Buret people arriving from Central Asia.

One interesting I observed through Genetikers phenotype analysis , is Iran_N peoples shared traits more akin to those found in MA-1 , ie dark brown skin, dark eyes and black hair
https://genetiker.wordpress.com/2016/07/18/phenotype-snps-from-ancient-iran/

CHG on the other had the derived SLC24A5 gene and were relatively light skinned.

The so called Caucasus is mainly coming Neolithic Iranians who arrived in South Asia 10 kya and/or the demic diffusion of mixed Steppe and Central Asian/BMAC agriculturalist(also mainly Iran_N descended) populations that diffused across NW South Asia the in the Bronze age.

A few points:

As you say derived SLC24A5 was present in CHG. But it is also present in about 50% on the Singhalese who are about as dark colored as any on the Subcontinent. SLC45A2 on the other hand is indeed a distinguishing feature.

Also: "Derived SLC24A5 rs1426654 alleles can also be observed in the AH1, AH4, and F38 individuals" AH1 and AH4 are Neolithic Iranians. http://science.sciencemag.org/content/sci/suppl/2016/07/13/science.aaf7943.DC1/Broushaki.SM.pdf

Another issue, while WC1 looks closest to modern populations from the subcontinent, we also see this from formal stats: "D-statistics show that Kotias is a better surrogate for Ancestral North Indians than GD13a ... For all modern southern Asian populations we tested, Kotias was a better putative source than GD13a (Fig. S9, Table S6)." http://biorxiv.org/content/biorxiv/suppl/2016/06/18/059568.DC1/059568-1.pdf

https://d2ufo47lrtsv5s.cloudfront.net/content/sci/early/2016/07/13/science.aaf7943/F4.large.jpg?width=800&height=600&carousel=1

pegasus
01-02-2017, 06:08 PM
A few points:

As you say derived SLC24A5 was present in CHG. But it is also present in about 50% on the Singhalese who are about as dark colored as any on the Subcontinent. SLC45A2 on the other hand is indeed a distinguishing feature.

Also: "Derived SLC24A5 rs1426654 alleles can also be observed in the AH1, AH4, and F38 individuals" AH1 and AH4 are Neolithic Iranians. http://science.sciencemag.org/content/sci/suppl/2016/07/13/science.aaf7943.DC1/Broushaki.SM.pdf

Another issue, while WC1 looks closest to modern populations from the subcontinent, we also see this from formal stats: "D-statistics show that Kotias is a better surrogate for Ancestral North Indians than GD13a ... For all modern southern Asian populations we tested, Kotias was a better putative source than GD13a (Fig. S9, Table S6)." http://biorxiv.org/content/biorxiv/suppl/2016/06/18/059568.DC1/059568-1.pdf

https://d2ufo47lrtsv5s.cloudfront.net/content/sci/early/2016/07/13/science.aaf7943/F4.large.jpg?width=800&height=600&carousel=1


1. Singhalese are modern day peoples, I was comparing 2 ancient populations so your point is moot.

2. That paper came out before more indepth papers on Iran_N populations came out. Clearly Steppe admixture is very low to nil in most Central/South Indian populations as they seem to be mainly 2 way mixes between Neolithic Iranians and local Paniya like foragers. Also your taking those papers out of context in regards to SLC45A2.
I do not posit too much in D stats, some users have modelled Pania as 40% Iran_N (thankfully he was banned) and Kurds as having as much Steppe and Iran_N dna as some SC Asian populations, both of which are ridiculous.

parasar
01-02-2017, 08:09 PM
1. Singhalese are modern day peoples, I was comparing 2 ancient populations so your point is moot.

2. That paper came out before more indepth papers on Iran_N populations came out. Clearly Steppe admixture is very low to nil in most Central/South Indian populations as they seem to be mainly 2 way mixes between Neolithic Iranians and local Paniya like foragers. Also your taking those papers out of context in regards to SLC45A2.
I do not posit too much in D stats, some users have modelled Pania as 40% Iran_N and Kurds as having as much Steppe and Iran_N dna as some SC Asian populations, both of which are ridiculous.

No it is not, as Iran Neolithic samples AH1 and AH4 are derived at SLC24A5 too.

Plus I did not bring up SLC45A2 in context of those papers. I just pointed out that SLC45A2 is the real distinguishing marker for skin color, not SLC24A5.

Why do you think Kurds can't have as much steppe and Iran N as some SC Asian populations?
Any why can't the Paniya can't have a 40% overlap with a sample like WC1?

Magnetic
01-02-2017, 08:28 PM
Hey all,

Could someone help me break down what it means that a lot of those who are identified as South Asian see tons of Caucasian in their admixtures?

I'm reading there is some Kurdish, or Turkic in us. What's the current data on the topic?

I know my Pakstani/Indian history but not really anything on the "other" part of my admixture. I should add that some of the stuff that others talk about goes over my head. If someone could break down to layperson terms.

you dont have Kurdish or Turkish in you . it is just Iran_Neolithic component

Awale
01-02-2017, 10:03 PM
As you say derived SLC24A5 was present in CHG. But it is also present in about 50% on the Singhalese who are about as dark colored as any on the Subcontinent. SLC45A2 on the other hand is indeed a distinguishing feature.

If a population has reached near fixation with SLC24A5's derived A allele at the rs1426654 SNP (like with Europeans) then you're essentially looking at 25-40% the pigmentation difference between Europeans and Afro-Carribs. Or, roughly an average 10 point drop in the Melanin Index (http://oi68.tinypic.com/1zo8cnk.jpg)... 50% derived is not that much though and they may even be like Somalis where we're 60% derived (https://alfred.med.yale.edu/alfred/SiteTable1A_working.asp?siteuid=SI007419V) but many are AG instead of being AA. Nevertheless, being 50% derived is interesting given their substantive non-West-Eurasian ancestry; implies their West-Eurasian predecessors were quite derived, in humble opinion. And, perhaps similar to what I and some others think regarding the ancestors of Horners and the North-Sudanese, the Sinhalese' mixed ancestors may very well have selected for further pigmentation than the initial mixing produced given their environment:

http://oi67.tinypic.com/2qjjhat.jpg

Effectively losing the derived G allele at SLC45A2's rs16891982 SNP as well, which does seem to be the case with Horn Africans (especially the likes of Beta Israels and Habeshas who lack the allele but definitely had Yemenite-Jew/Copt-like ancestors (see here (http://anthromadness.blogspot.ae/2016/01/the-later-west-eurasian-element-in-agaws.html) and here (http://anthromadness.blogspot.ae/2016/05/using-somalis-as-proxy-second-attempt.html)) who most likely carried it while Somalis possibly did have Levant_N-like ancestors who carried it as well).

I recommend reading this study: http://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1003912

pegasus
01-02-2017, 10:16 PM
No it is not, as Iran Neolithic samples AH1 and AH4 are derived at SLC24A5 too.

Plus I did not bring up SLC45A2 in context of those papers. I just pointed out that SLC45A2 is the real distinguishing marker for skin color, not SLC24A5.

Why do you think Kurds can't have as much steppe and Iran N as some SC Asian populations?
Any why can't the Paniya can't have a 40% overlap with a sample like WC1?

I think thats pretty self explanatory . Again ur taking things out of context. Pania cannot be modelled as 40% Iran_N , land Parasar its a clear day.

Another tidbit , in regards to Kotias. CHG like ancestry in South Asians came with Steppe populations not Chalcolithic Iranians, because they lack the Barcin Anatolian rich ancestry you see in many West Asians. So that creates a paradox you have huge swathes of populations , which do have low or nil steppe ancestry but do have significant "ANI" ancestry , that clearly is coming from Iran_N peoples.

parasar
01-02-2017, 10:53 PM
If a population has reached near fixation with SLC24A5's derived A allele at the rs1426654 SNP (like with Europeans) then you're essentially looking at 25-40% the pigmentation difference between Europeans and Afro-Carribs. Or, roughly an average 10 point drop in the Melanin Index (http://oi68.tinypic.com/1zo8cnk.jpg)... 50% derived is not that much though and they may even be like Somalis where we're 60% derived but many are AG instead of being AA. Nevertheless, being 50% derived is interesting given their substantive non-West-Eurasian ancestry; implies their West-Eurasian predecessors were quite derived, in humble opinion. And, perhaps similar to what I and some others think regarding the ancestors of Horners and the North-Sudanese, the Sinhalese' mixed ancestors may very well have selected for further pigmentation than the initial mixing produced given their environment:

http://oi67.tinypic.com/2qjjhat.jpg

Effectively losing the derived C allele at SLC45A2's rs16891982 SNP as well, which does seem to be the case with Horn Africans (especially the likes of Beta Israels and Habeshas who lack the allele but definitely had Yemenite-Jew/Copt-like ancestors (see here (http://anthromadness.blogspot.ae/2016/01/the-later-west-eurasian-element-in-agaws.html) and here (http://anthromadness.blogspot.ae/2016/05/using-somalis-as-proxy-second-attempt.html)) who most likely carried it while Somalis possibly did have Levant_N-like ancestors who carried it as well).

I recommend reading this study: http://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1003912

Yes I have read that one on IBD.
Another one making a similar point:
http://www.g3journal.org/content/3/11/2059.full
"Virtually all chromosomes carrying the A111T allele share a single 78-kb haplotype that we call C11, indicating that all instances of this mutation in human populations share a common origin. The C11 haplotype was most likely created by a crossover between two haplotypes, followed by the A111T mutation. The two parental precursor haplotypes are found from East Asia to the Americas but are nearly absent in Africa ... To learn about the time and location of origin of the A111T mutation, we studied haplotypes in the region around SLC24A5 across world populations ... the greatest diversity of B-region haplotypes associated with C11 is found in GIH (89% B6)."

Awale
01-03-2017, 11:12 PM
Why do you think Kurds can't have as much steppe and Iran N as some SC Asian populations?

Hmm, regarding just their steppe ancestry... I find it hard to believe that they're remotely as steppic as some SC Asians (i.e. Afghan Pashtuns). Kurds have always just seemed barely distinguishable from Plateau Iranians like various Persian speakers... Only perhaps sometimes more Near Eastern-shifted versions of said Plateau Iranians (which makes sense geographically):

http://oi67.tinypic.com/2vjzlm1.jpg
http://oi63.tinypic.com/okwxms.jpg

I mean look at how they cluster in respect to Iran_Chls and ELBA steppe populations:

http://oi68.tinypic.com/1hwazb.jpg

I'd say Kurds are probably pretty indigenous to the Near East overall.

Kurd
01-04-2017, 03:03 AM
Hmm, regarding just their steppe ancestry... I find it hard to believe that they're remotely as steppic as some SC Asians (i.e. Afghan Pashtuns). Kurds have always just seemed barely distinguishable from Plateau Iranians like various Persian speakers... Only perhaps sometimes more Near Eastern-shifted versions of said Plateau Iranians (which makes sense geographically):

http://oi67.tinypic.com/2vjzlm1.jpg
http://oi63.tinypic.com/okwxms.jpg

I mean look at how they cluster in respect to Iran_Chls and ELBA steppe populations:

http://oi68.tinypic.com/1hwazb.jpg

I'd say Kurds are probably pretty indigenous to the Near East overall.



I am not seeing what you are seeing....the distance from the center of the cluster marked "Kurdish" to the 2 Andronovo samples is actually less than the distance from most of the SC Asian samples (grey Xs) to the 2 Andronovo samples.

This said, a PCA is actually not the best tool for figuring out how much steppe ancestry groups such as Kurds and Pashtuns have, for a couple of reasons:

- The 2 dimensions, PC1 and PC2 can't capture all the genetic variation in the samples;
- If the PCA is ADMIXTURE based then the output changes every time you change the number or type of samples in the dataset, and/or the component references. ADMIXTURE is driven by allele sharing at a few positions, resulting from recent interactions between individuals. It is no coincidence that an ADMIXTURE based PCA mirrors geographical location of the samples. Thus, expectedly Kurds cluster with Iranians, and Pashtuns cluster with other SC Asians.


The major papers use formal stats which directly compare 2 samples or 2 pops to gauge shared drift between them. The dstats I posted at http://www.anthrogenica.com/showthread.php?8898-Excess-Steppe-Ancestry-in-Kurds, which directly compare Kurds, W Asians, and SC Asians with steppe pops show that Kurds share more drift with pretty much all the steppe groups than most of their neighbors, and many SC Asians. This likely reflects ancient genetic input originating in Pontic-Caspian steppe and Central Asia, something that is not accurately picked up with ADMIXTURE.


Just for kicks, I will run IBD comparisons later between Kurds, W Asians, and SC Asians, with steppe groups.

Kurd
01-04-2017, 03:09 AM
you dont have Kurdish or Turkish in you . it is just Iran_Neolithic component

Although the Iran N samples recovered from Kurdistan showed that they form a large chunk of the ancestry of the various Kurd, Iranian, SC Asian, and S Asian tribes, the probability that gene flow from the Kurdistan or Iran area into SC/ S Asia ceased post neolithic is probably near 0. Case in point groups such as Baloch and Brahui migrating E from the NW Iran area post neolithic.

pegasus
01-04-2017, 06:15 AM
Although the Iran N samples recovered from Kurdistan showed that they form a large chunk of the ancestry of the various Kurd, Iranian, SC Asian, and S Asian tribes, the probability that gene flow from the Kurdistan or Iran area into SC/ S Asia ceased post neolithic is probably near 0. Case in point groups such as Baloch and Brahui migrating E from the NW Iran area post neolithic.

The Baloch language came from NW Iran within the past 1000 years, by that date most people in that region are mostly indistinguishable from the ones today, even that 14th-15th century Iranian they found from the Zagros region was indistinguishable from modern Iranians. I am quite appalled that even in light of the papers from last year on Zagros Farmers your still harping on the logic that Baloch/Brahui are somehow Diaspora Kurds. Brahui speak a Dravidian language for the same amount of time, does that make them South Indians? Clearly modern day Baloch and Brahui are mainly descended from peoples who have been living there for a very long time and they seem to very much connected with populations living in Iran's coastal areas in the deep south like Hormozgan who also have been living in the region for millenia. They are not from NW Iran and are not Kurds. You blatantly turn a blind eye to these facts on one hand and on the other hand claim Kurdistan is some bastion of Steppe ancestry ( more so than SC Asians and even some South Asians) is beyond the pale , given the fact Indo Iranian groups have a prolonged history in SC Asia starting in the Bronze age and continuing all the way into late antiquity. That region has witnessed an unbelievable onslaught from these people.

Awale is right Kurds are very much a Near Eastern/West Asian people and are not remotely steppic as SC Asians.

Kurd
01-04-2017, 11:15 AM
The Baloch language came from NW Iran within the past 1000 years

Good..., so you agree that the Baloch have migrated from NW Iran relatively recently, and are thus genetically closely related to the Kurds living there. :)


Awale is right Kurds are very much a Near Eastern/West Asian people


and Pashtuns are located in SC Asia, and Punjabis in S Asia...........what does any of this have to do with how related they are to steppe populations, and how much ancestry they draw from the steppes?


and are not remotely steppic as SC Asians.


You make alot of claims but post no evidence......how about posting some evidence like I did for change, I promise I will take you more seriously if you do :) . Explain to me the following in this case:

1- Why does the PCA Awale posted show Kurds closer to Andronovo, than most SC Asians to Andronovo?
2- Why are the dstats I posted at http://www.anthrogenica.com/showthread.php?8898-Excess-Steppe-Ancestry-in-Kurds , D ( SC Asians, Kurd, Steppe, Outgroup) negative for many SC Asians, AND at the same time D ( W Asians, Kurd, Steppe, Outgroup) negative for most W Asians ?


Not saying that Kurds are more related to steppe groups than ALL SC Asians. My own analysis has shown that various Tajik groups are for example more closely related to steppe populations than Kurds are to steppe populations, however, we can not generalize and say that all SC Asians are more closely related to steppe groups than any Kurd tribe is related to steppe groups, based on the evidence

Coldmountains
01-04-2017, 03:52 PM
Good..., so you agree that the Baloch have migrated from NW Iran relatively recently, and are thus genetically closely related to the Kurds living there. :)



and Pashtuns are located in SC Asia, and Punjabis in S Asia...........what does any of this have to do with how related they are to steppe populations, and how much ancestry they draw from the steppes?




You make alot of claims but post no evidence......how about posting some evidence like I did for change, I promise I will take you more seriously if you do :) . Explain to me the following in this case:

1- Why does the PCA Awale posted show Kurds closer to Andronovo, than most SC Asians to Andronovo?
2- Why are the dstats I posted at http://www.anthrogenica.com/showthread.php?8898-Excess-Steppe-Ancestry-in-Kurds , D ( SC Asians, Kurd, Steppe, Outgroup) negative for many SC Asians, AND at the same time D ( W Asians, Kurd, Steppe, Outgroup) negative for most W Asians ?


Not saying that Kurds are more related to steppe groups than ALL SC Asians. My own analysis has shown that various Tajik groups are for example more closely related to steppe populations than Kurds are to steppe populations, however, we can not generalize and say that all SC Asians are more closely related to steppe groups than any Kurd tribe is related to steppe groups, based on the evidence

It is easy to see that Baluchs are not recent kurdish immigrants because their Y-DNA is very much local and closest to Sindhi, Pashtuns, eastern Persians and Brahui. They are rich in R1a-L657 for example and i not heard of one Kurd belonging to this lineage. They may have some recent western iranic ancestry but they pretty much fit in the region just less steppe and more Iran_N shifted. Kurds are low in R1a (20%) and some groups have even less than 10% of it. That is not a proof that they have less steppe ancestry but most steppe ancestry was mediated by males so it is unlikely that they have almost as much as steppe ancestry as Pashtuns or NW Indians. Pashtuns, Tajiks, NW Indians (high caste+ Jatts) have definitely more steppe ancestry and they have also much more NE Euro-like ancestry. Kurds have maybe almost as much or more steppe ancestry than some Baluch or very east-asian shifted turkic Central Asian but Indo-Iranians arrived much earlier in South Central Asia and this region had several immigration waves of Indo-Iranians (Proto-Nuristani, Proto-Indo-Aryans, Proto-Iranics, Kushan, Saka,...). Kurdistan was just populated by West Iranics and later some Saka settled there but their number was very much limited

Magnetic
01-04-2017, 05:39 PM
we are a west asian population and our autosomal dna is very similar to other west asian populations . we cluster with other west asians . not sure what some people are smoking but it is getting annoying

pegasus
01-04-2017, 06:17 PM
Good..., so you agree that the Baloch have migrated from NW Iran relatively recently, and are thus genetically closely related to the Kurds living there. :)



and Pashtuns are located in SC Asia, and Punjabis in S Asia...........what does any of this have to do with how related they are to steppe populations, and how much ancestry they draw from the steppes?




You make alot of claims but post no evidence......how about posting some evidence like I did for change, I promise I will take you more seriously if you do :) . Explain to me the following in this case:

1- Why does the PCA Awale posted show Kurds closer to Andronovo, than most SC Asians to Andronovo?
2- Why are the dstats I posted at http://www.anthrogenica.com/showthread.php?8898-Excess-Steppe-Ancestry-in-Kurds , D ( SC Asians, Kurd, Steppe, Outgroup) negative for many SC Asians, AND at the same time D ( W Asians, Kurd, Steppe, Outgroup) negative for most W Asians ?


Not saying that Kurds are more related to steppe groups than ALL SC Asians. My own analysis has shown that various Tajik groups are for example more closely related to steppe populations than Kurds are to steppe populations, however, we can not generalize and say that all SC Asians are more closely related to steppe groups than any Kurd tribe is related to steppe groups, based on the evidence

Your so called evidence and analysis often times is doctored/manipulated results which skew the facts don't take my word for it there are many members who have noticed gaping discrepencies . In another thread both DMXX and Generalissimo had to call you out on your methodologies. The facts are out based of published papers . Reasoning with blatant ignorance is difficult.

Also don't put words in my mouth. I never said Baloch/Brahui are Kurds, they clearly are not. By your logic every African American should jump up and down claim to be Anglo Saxon because they have 10-20% Northern European ancestry. As a Pakistani man you seem more obsessed with being Kurdish than the actual Kurdish members on the forum.

The fact the earliest neolithic sites in South Asia are in Balochistan and you have the same Neolithic ancestry peaking in the region is no coincidence. Sometimes its not hard putting 2 and 2 together.
Interestingly even the Sindhis who have posted their results plot much closer to Baloch/Brahui than any Kurd .

Awale
01-04-2017, 08:02 PM
This said, a PCA is actually not the best tool for figuring out how much steppe ancestry groups such as Kurds and Pashtuns have, for a couple of reasons:

- The 2 dimensions, PC1 and PC2 can't capture all the genetic variation in the samples;
- If the PCA is ADMIXTURE based then the output changes every time you change the number or type of samples in the dataset, and/or the component references. ADMIXTURE is driven by allele sharing at a few positions, resulting from recent interactions between individuals. It is no coincidence that an ADMIXTURE based PCA mirrors geographical location of the samples. Thus, expectedly Kurds cluster with Iranians, and Pashtuns cluster with other SC Asians.

Interesting criticisms but none of those PCAs were ADMIXTURE based and the tool itself is honestly a pretty great one when you know how to use it and can properly suss out the various affinities populations show for each other (based on how they're clustering in respect to those dimensions) which, quite often, tend to correlate well with their ADMIXTURE and formal stat results.

In this case, it's very straight-forward. The more steppic a population is; the closer you would expect them to cluster to ELBA steppe populations. Now, Kurds don't pull any closer to ELBA steppe populations than the Iranian or Azeri samples do. In fact, those Kurd samples cluster extremely close to the two aforementioned populations (sometimes even overlapping with them). Now that says something quite clear about their affinities toward populations like the Sintashta, Andronovo, Yamnaya, Srubnaya, and so on... It says they're clearly barely distinct from the Iranian and Azeri lot in this respect which correlates with most ADMIX and formal stat results I've seen, other than your own. And which basically imparts that they're not as steppe influenced as populations like Afghan Pashtuns. In fact, despite their ASI-related ancestry (or other such non-West Eurasian affinities) probably skewing their clustering to some extent; some SC Asians clearly still pull more toward ELBA steppe populations than Kurds do:

http://oi68.tinypic.com/zxmmj5.jpg


The major papers use formal stats which directly compare 2 samples or 2 pops to gauge shared drift between them. The dstats I posted at http://www.anthrogenica.com/showthre...estry-in-Kurds, which directly compare Kurds, W Asians, and SC Asians with steppe pops show that Kurds share more drift with pretty much all the steppe groups than most of their neighbors, and many SC Asians. This likely reflects ancient genetic input originating in Pontic-Caspian steppe and Central Asia, something that is not accurately picked up with ADMIXTURE.

I have a hard time trusting your analyses here, my compadre. They never seem to correlate well with what other sources tend to posit with their own formal stats or ADMIX runs or whatnot where Kurds barely seem distinct from some other West-Iranic speakers (see here (https://docs.google.com/spreadsheets/d/1tFAa7oxWpcNN-OdMMjBdb4NeWKG7EkpKMzZJVW2_MME/edit#gid=1330589433) and compare them to the likes of the Iranian-Lur and Iranian-Persian samples, for instance). Your recent encounter with David, DMXX and others is a good example of what I mean as well...


Good..., so you agree that the Baloch have migrated from NW Iran relatively recently, and are thus genetically closely related to the Kurds living there. :)

No, I believe he was saying that the language came from NW Iran, not the Baloch people as a whole who, like everyone has been saying, seem mostly "native" to SC Asia. The original speakers of their ancestral language didn't have to be genetically identical to them and, in all honestly, I'd be surprised if they were if they really did come from around Northwestern Iran of all places given the more Iran_ChL and lower steppe autosomal profile predominant there when compared to SC Asia.

--

I'll leave it at that. I can't say much else that other parties (Coldmountains, Pegasus and Magnetic) have not already said.

Kurd
01-05-2017, 01:03 AM
Your so called evidence and analysis often times is doctored/manipulated results which skew the facts

This is a serious charge that I don't take lightly. I hope you realize that I can easily bring a defamation of character law suit against you if I choose to. I have your post as evidence of the defamatory statement and the false statement of fact, to wit, "my results are doctored", and all I have to do is give my dataset to someone who knows and is able to run dstats, and show that they come up with the same results as me. As far as unveiling your identity and address, all I would have to do is subpoena this website's records from their domain host on the East Coast, and take my word for it I have alot of experience in this area.

In fact, you could be exposing this website to liability, since you have published a serious charge, without offering proof that my results are doctored.

Kurd
01-05-2017, 02:57 AM
Interesting criticisms but none of those PCAs were ADMIXTURE based and the tool itself is honestly a pretty great one when you know how to use it and can properly suss out the various affinities populations show for each other (based on how they're clustering in respect to those dimensions) which, quite often, tend to correlate well with their ADMIXTURE and formal stat results.

In this case, it's very straight-forward. The more steppic a population is; the closer you would expect them to cluster to ELBA steppe populations. Now, Kurds don't pull any closer to ELBA steppe populations than the Iranian or Azeri samples do. In fact, those Kurd samples cluster extremely close to the two aforementioned populations (sometimes even overlapping with them).



While on the PCA you posted, the SC Asians enclosed in the red circle are a little closer to Andronovo, the majority of the SC Asians outside the circle you drew are not closer to Andronovo than the Kurds are to Andronovo.

No one is saying that Kurds don't cluster with Iranians in ADMIXTURE or PCA, however, dstats unveil a hidden facet not visible to ADMIXTURE or PCA. Dstats basically unveil older mutations not picked up in ADMIXTURE (feel free to email Nick Patterson to ask him why there is an inconsistency between ADMIXTURE and dstats and whether dstats look at older mutation, and please tell us what he says). So basically, ADMIXTURE links Kurds and Iranians with a common denominator set of more recent mutations, whereas dstats show older mutations and gene flow. That is what the dstats I posted under the Kurdish thread pick up on, whether it be old gene flow into Kurds or Kurdistan from the N Caucausus, Mittani, Scythians, Hurrians, or what have you.

I can't help it if others are not posting detailed in depth dstats comparing Kurds with various other groups.



No, I believe he was saying that the language came from NW Iran, not the Baloch people as a whole


So if Baloch did not bring the language from NW Iran to Balochistan, then who did? Kurds? Ghosts? Whoever brought the language there surely did not leave their genes behind in NW Iran.... no?

Gravetto-Danubian
01-05-2017, 05:52 AM
Kurd
Can you or any other Kurd submit your personal data to Davidski to creat a GlobalK10 file ?
I'd like to put it though nMonte..

Kurd
01-05-2017, 07:28 AM
Kurd
Can you or any other Kurd submit your personal data to Davidski to creat a GlobalK10 file ?
I'd like to put it though nMonte..


I had submitted one of my northern Iraqi Kurd samples to him. This is what he sent:

PC1,PC2,PC3,PC4,PC5,PC6,PC7,PC8,PC9,PC10
Kurd_Iraq:Kurd_Iraq,0.0141,0.0224,-0.0012,0.0198,0.0075,-0.0237,-0.0093,0.0002,0.0208,-0.0023

Have at it. The K7 ADMIXTURE output showed it clustering with other Kurds and Iranians, which is expected. No idea what your nMonte will look like..

Gravetto-Danubian
01-05-2017, 08:46 AM
I had submitted one of my northern Iraqi Kurd samples to him. This is what he sent:

PC1,PC2,PC3,PC4,PC5,PC6,PC7,PC8,PC9,PC10
Kurd_Iraq:Kurd_Iraq,0.0141,0.0224,-0.0012,0.0198,0.0075,-0.0237,-0.0093,0.0002,0.0208,-0.0023

Have at it. The K7 ADMIXTURE output showed it clustering with other Kurds and Iranians, which is expected. No idea what your nMonte will look like..


I was curious to see if they have anything different, or to test whether Scythian e.a. ancestry can be picked up.
Seems like this individual is very much similar to Caucasus groups - esp Armenians, characterized by high Armenian EBA ancestry, extra Kotias, a markedly attenuated Iran Neolithic ancestry, and significant Levant Bronze Age admixture at 'K6000BC" level. Curiously, they do have more (double) EHG + some Dai c.f. Armenians.

13458

A visual. (not the optimal PCA, as would require weighting of related components, and addition of ancients and non-IE west Asians also).

13459

Let me know if you want another configuraiton tested (as in different age source Pops - as in Chalcolithic and more recent prehistory all included).

Kurd
01-05-2017, 12:13 PM
I was curious to see if they have anything different, or to test whether Scythian e.a. ancestry can be picked up.
Seems like this individual is very much similar to Caucasus groups - esp Armenians, characterized by high Armenian EBA ancestry, extra Kotias, a markedly attenuated Iran Neolithic ancestry, and significant Levant Bronze Age admixture at 'K6000BC" level. Curiously, they do have more (double) EHG + some Dai c.f. Armenians.

http://www.anthrogenica.com/attachment.php?attachmentid=13458&stc=1

A visual. (not the optimal PCA, as would require weighting of related components, and addition of ancients and non-IE west Asians also).

http://www.anthrogenica.com/attachment.php?attachmentid=13459&stc=1

Let me know if you want another configuraiton tested (as in different age source Pops - as in Chalcolithic and more recent prehistory all included).



Just a couple of questions:

1- what is Mentese;
2- whenever you have time can you run a few of these without Kotias, since Iran N and Kotias have substantial overlap
3- I'd also be curious to see what happens under scenario 2, when Scythian is substituted for Samara as the Steppe proxy

Gravetto-Danubian
01-05-2017, 02:05 PM
Just a couple of questions:

1- what is Mentese;

Anatolian Neolithic from Mentese averaged out. Its essentially the same as Barcin. All early Anatolian farmers (=ANF), except one of the Tepecik's, are homogeneous.



2- whenever you have time can you run a few of these without Kotias, since Iran N and Kotias have substantial overlap

Yes and no. Iran Neolithic differs from Kotias in the exact proportions of 'basal' and archaic ANE. Plus Iran Neolithic has Onge/ Andamanese like ancestry (also representable by ust-ishm) which Kotias and Satsurblia do not .
Moreover, IMO, nMOnte can correctly "choose" between these two otherwise very similar ancestral components, even when presented with both. In this regard, nMonte can act as an objective 'unsupervised' test.
So what the result above shows is that all the 'Caucasus' type ancestry in Indians comes from Iran Neolithic, whislt Armenians and Kurds have extra additional Kotias-like ancestry from some other source: the results I will post tomorrow will show that most of this is represented by "EBA Armenia" (i.e. the Kura -Arax expansion). Furthermore, Armenians (& other Caucasians) have even extra CHG over and above EBA Armenia; i.e. actual Mesolithic Kotias relics, which is what some papers' and others' IBD analysis were picking up. But to answer your question, removing Kotias drives up Iran Neol by almost 1 for 1 % points, with some minor additional rise in EHG too (to compensate).


3- I'd also be curious to see what happens under scenario 2, when Scythian is substituted for Samara as the Steppe proxy

There are many ways we can play with the modelling, to unravel layers of ancestry, epoch by epoch. I'll elaborate tomorrow

Kurd
01-05-2017, 05:15 PM
A couple of odd things that stand out is that Kurd Iraq shows 0 Mentese ANF whereas the Iranian samples show copious amounts, considering Kurd's genetic links with the Anatolia region. The other is the high Levant N which I have not been able to substantiate using formal methods.

Fortunately, we have simple and reliable tools such as Dstats that I can use to a direct one to one comparison to verify the above results D(Ancient1, Ancient2, Kurd, Outgroup) will tell us whether Kurd shares more alleles with Ancient1 or Ancient2, and we can go through the various combinations of ancients, such as ANF-Levant N, Kotias-LevantN, etc. I will try to get to it tonight.

The other tool I have at my disposal is Beagle, and I can run Kurd against the various ancients for IBD, and total out the segments, after pruning all the ancients to the same number of SNPs of course.

These tests will tell us whether the nMonte output is reasonable.

Kurd Iraq having about 3.5X the amount of EHG as Armenian and more Dai in your nMonte is not surprising based on my formal test results.

Although nMonte appears to be a useful tool, I don't use it. I commented briefly on it at http://www.anthrogenica.com/showthread.php?9181-Davidski-s-Basal-rick-K7-Global-10-Genetic-Map-Results&p=201421#post201421

Look forward to your outputs.

Awale
01-06-2017, 12:29 AM
While on the PCA you posted, the SC Asians enclosed in the red circle are a little closer to Andronovo, the majority of the SC Asians outside the circle you drew are not closer to Andronovo than the Kurds are to Andronovo.

You should re-read what I wrote:


In fact, despite their ASI-related ancestry (or other such non-West Eurasian affinities) probably skewing their clustering to some extent; some SC Asians clearly still pull more toward ELBA steppe populations than Kurds do:

That regional PCA can sometimes be pretty uninformative if a population isn't at least 80%+ West-Eurasian (an ANE, Basal-rich and Villabruna-related mixture of some sort) and some of those populations are probably being skewed in their clustering by their varying degrees of non-West Eurasian ancestry (ASI et al.). But who knows... Still stands that a good chunk of them (whom I encircled) pull closer to Steppe populations than Kurds do but, honestly, I mentioned this in passing... The main dilemma is indeed, for the most part, that Kurds in all analyses that I've seen (other than your own) don't look very distinct from Azeris, Lurs, Iranian Plateau Persians, and such and such (all populations we know aren't more steppic than SC Asians like Afghan Pashtuns).


No one is saying that Kurds don't cluster with Iranians in ADMIXTURE or PCA, however, dstats unveil a hidden facet not visible to ADMIXTURE or PCA. Dstats basically unveil older mutations not picked up in ADMIXTURE (feel free to email Nick Patterson to ask him why there is an inconsistency between ADMIXTURE and dstats and whether dstats look at older mutation, and please tell us what he says). So basically, ADMIXTURE links Kurds and Iranians with a common denominator set of more recent mutations, whereas dstats show older mutations and gene flow. That is what the dstats I posted under the Kurdish thread pick up on, whether it be old gene flow into Kurds or Kurdistan from the N Caucausus, Mittani, Scythians, Hurrians, or what have you.

Right, and the only stats I've seen or heard of that imply what you're saying here, as others have noted (http://www.anthrogenica.com/showthread.php?7560-Lazaridis-et-al-(2016)-A-look-at-various-methods-including-qpAdm-Dstats-f3&p=202333#post202333), seem to be your own. You see the dilemma? Granted, and no offense intended, I trust David somewhat more than I trust you. He's been at this for ages and knows some of these analyses better than most academics whilst you seem newer to all of this and not as sharpened. And, honestly, even without d-stats or qpAdm runs from other people like David; one should be suspicious when one mode of analysis is painting a completely distinct picture from all others (especially when there's clearly nothing wrong with the other analyses). It's pretty unscientific to look at 2-3 analyses that imply a population isn't very steppic at all (10-20% at best?) and then see one analysis that implies MUCH more of such admixture and immediately jump on that one analysis.


So if Baloch did not bring the language from NW Iran to Balochistan, then who did? Kurds? Ghosts? Whoever brought the language there surely did not leave their genes behind in NW Iran.... no?

... Are you 1:1-ing languages and peoples/genomics or something? You realize the Baloch, as they are now (genomically and such), did not have to have migrated from NW Iran to have their language, yes? It could've easily just been a different people from NW Iran (perhaps folk who were pretty genomically similar to Lurs, Kurds, and Iranian Plateau Persians) who migrated over and caused the local non-Baloch speaking ancestors of the modern Baloch to, for whatever reasons, eventually language-shift whilst maybe imparting not much ancestry to them at all in the long-run. Essentially keeping the "Baloch" mostly "local" from a genomic standpoint which is indeed what they look to be.

khanabadoshi
01-06-2017, 03:38 AM
I was curious to see if they have anything different, or to test whether Scythian e.a. ancestry can be picked up.
Seems like this individual is very much similar to Caucasus groups - esp Armenians, characterized by high Armenian EBA ancestry, extra Kotias, a markedly attenuated Iran Neolithic ancestry, and significant Levant Bronze Age admixture at 'K6000BC" level. Curiously, they do have more (double) EHG + some Dai c.f. Armenians.



A visual. (not the optimal PCA, as would require weighting of related components, and addition of ancients and non-IE west Asians also).



Let me know if you want another configuraiton tested (as in different age source Pops - as in Chalcolithic and more recent prehistory all included).

Would you mind posting your nMonte population spreadsheet? I'd like to run my family and the other member samples I have from the global 10. Your population choices seem well thought out; curious to see how they compare to what you've already posted.

I will probably check back in a few days. Been busy as of late. Thanks for posting, or considering so, in advance.

Kurd
01-06-2017, 04:00 AM
And, honestly, even without d-stats or qpAdm runs from other people like David; one should be suspicious when one mode of analysis is painting a completely distinct picture from all others (especially when there's clearly nothing wrong with the other analyses). It's pretty unscientific to look at 2-3 analyses that imply a population isn't very steppic at all (10-20% at best?) and then see one analysis that implies MUCH more of such admixture and immediately jump on that one analysis.



Lets face it the predominant analysis you and many others have seen are ADMIXTURE outputs and PCAs. The outputs from both of these are highly volatile, and dependent on which populations you include in the dataset, and in the case of ADMIXTURE, whether the run is supervised or not, and which samples we ADMIXTURE calculator creators declare ancestral. I have personally observed this over and over again, in all the specialized calculators I have created whether on gedmatch or DIYs. As a calculator creator I reluctantly say this, but do in the interest of being fair.

That is why us calculator creators, whether it be me or David spend hours, and sometimes days adding and subtracting samples from the runs, changing component references, and K, until the results seem somewhat consistent with formal tests, and NOT the other way around. I did this for my ANE K6, and am sure David did the same for his latest K7. So please stop worshiping ADMIXTURE calculators and PCAs, and please do me and yourself a favor by concentrating more on formal methods, especially dstats which do direct sample to sample analysis for derived allele sharing (unlike ADMIXTURE and PCA, output is stable, no matter which samples are in the dataset).

Here is what David also cautioned lately, so it will be beneficial to you and everyone else to spend a little more time studying formal tests, and reading N Patterson's paper. In fact even better, if you learn how to do formal tests, you can yourself verify everything I post.

"I wouldn't rely too much on the K7, or indeed any ADMIXTURE output, to produce valid models." http://www.anthrogenica.com/showthread.php?9291-An-nMonte-and-4mix-Guide-for-the-Participants-of-the-Basal-rich-K7-and-or-Global-10-T&p=205940#post205940 (http://www.anthrogenica.com/showthread.php?9291-An-nMonte-and-4mix-Guide-for-the-Participants-of-the-Basal-rich-K7-and-or-Global-10-T&p=205940#post205940)



Right, and the only stats I've seen or heard of that imply what you're saying here, as others have noted (http://www.anthrogenica.com/showthread.php?7560-Lazaridis-et-al-%282016%29-A-look-at-various-methods-including-qpAdm-Dstats-f3&p=202333#post202333), seem to be your own. You see the dilemma?

The real dilemma here is that there is less than a handful of us that know how to run formal tests, and I am the only one who has dedicated his time exclusively to the study of S/SC/W Asian population histories. Therein lies the problem. There is not much to compare with my specific tests, however, you can always have David run tests that compare other W Asians with Kurd Iraq for shared drift with steppe (he has the Kurd Iraq sample). For example you can have him run D ( W Asians, Kurd Iraq, Steppe, Chimp) where W Asian is allowed to vary, and Steppe can be variable.

The same applies to the IBD runs I do with Beagle. I have not seen anyone else here posting the results from Beagle IBD runs they have done, never mind IBD comparisons involving W/SC/S Asians. So again, nothing to compare with. However, again, IBD would take precedence over ADMIXTURE, for all the reasons I mentioned above, plus some. I encourage you and others to learn how to perform these tests.


Are you 1:1-ing languages and peoples/genomics or something? You realize the Baloch, as they are now (genomically and such), did not have to have migrated from NW Iran to have their language, yes? It could've easily just been a different people from NW Iran (perhaps folk who were pretty genomically similar to Lurs, Kurds, and Iranian Plateau Persians) who migrated over and caused the local non-Baloch speaking ancestors of the modern Baloch to, for whatever reasons, eventually language-shift whilst maybe imparting not much ancestry to them at all in the long-run. Essentially keeping the "Baloch" mostly "local" from a genomic standpoint which is indeed what they look to be.





No guesswork needed here at all. The answer is fairly obvious when you run D( S Asian, Kurd, Baloch, Chimp). I have posted many results, but again you can have David run this test.

Kurd
01-06-2017, 04:36 AM
A couple of odd things that stand out is that Kurd Iraq shows 0 Mentese ANF whereas the Iranian samples show copious amounts, considering Kurd's genetic links with the Anatolia region. The other is the high Levant N which I have not been able to substantiate using formal methods.

Fortunately, we have simple and reliable tools such as Dstats that I can use to a direct one to one comparison to verify the above results D(Ancient1, Ancient2, Kurd, Outgroup) will tell us whether Kurd shares more alleles with Ancient1 or Ancient2, and we can go through the various combinations of ancients, such as ANF-Levant N, Kotias-LevantN, etc. I will try to get to it tonight.

The other tool I have at my disposal is Beagle, and I can run Kurd against the various ancients for IBD, and total out the segments, after pruning all the ancients to the same number of SNPs of course.

These tests will tell us whether the nMonte output is reasonable.

Kurd Iraq having about 3.5X the amount of EHG as Armenian and more Dai in your nMonte is not surprising based on my formal test results.

Although nMonte appears to be a useful tool, I don't use it. I commented briefly on it at http://www.anthrogenica.com/showthre...421#post201421

Ok, there is soime good news and some bad with regards to the what I assume is PCA based nMonte (?) results you posted. First the good, you are correct Kurd Iraq shares significantly more alleles/drift with Armenia EBA than with either CHG or Levant N:

Armenia_EBA CHG .Kurd_Iraq Chimp 0.0233 4.809 14374 13721 197910
evant_N Armenia_EBA .Kurd_Iraq Chimp -0.0281 -5.021 10581 11193 150850

Also, Kurd Iraq shares more alleles with CHG than with Iran N:

Iran_N CHG .Kurd_Iraq Chimp -0.0154 -2.489 11715 12082 166399

however, the reason I believe this is the case is because Kurd_Iraq is significantly more EHG shifted than Iran N, and as you know CHG is more EHG shifted than Iran N. A clue to this we see in the following, and here is the bad news:

.Kurd_Iraq Iran_Zoroastrian Iran_N Chimp 0.0008 0.110 3850 3844 52440
.Kurd_Iraq Iran_Fars Iran_N Chimp 0.0018 0.260 3865 3851 52440

Your nMonte output shows Iran Fars and Iran Zoro scoring much higher Iran N than Kurd Iraq, however the above dstats do not support those results, they show that Kurd Iraq shares as many alleles with Iran N ( if not insignificantly more) as Iran Fars and Iran Zoro.

Here is another problem with the nMonte output. As I suspected, Kurd Iraq showing 0 Anatolia N did not seem right, because of geography and the fact that all my Kurd samples show some type of affinity to Anatolia N. Dstats confirm this:

Anatolia_N CHG .Kurd_Iraq Chimp 0.0176 4.282 16341 15775 220868
Levant_N Anatolia_N .Kurd_Iraq Chimp -0.0246 -5.165 10636 11173 157233
.Kurd_Iraq Iran_Fars Anatolia_N Chimp 0.0080 1.862 5169 5088 68168
Kurd_Iraq Iran_Zoroastrian Anatolia_N Chimp 0.0058 1.338 5151 5092 68168


Also, the high Levant N score compared to others for Kurd in nMonte did not seem right. Although, Kurd does share drift with Levant N, Kurd shares more drift with CHG than with Levant N. nMonte shows the opposite.

evant_N CHG .Kurd_Iraq Chimp -0.0041 -0.678 11594 11689 157296

If those coordinates are PCA based, then that may explain the inconsistencies because PCA has some volatility where the output is dependent on the input pops.

Gravetto-Danubian
01-06-2017, 07:32 AM
Kurd e.a.

Here is a full approach to south & west Asians.

#1) northeastern West Asians (iranians, Kurds, Armenians) and South Asians can be (for the most part) modelled as a product of a 'Big 4' Ancestral groups: MA-1, Natufians, Ust-Ishm (stands in nicely for Onge or Paniya, can use either one), and Dai (for "east Asian")

13488

You will see that Satsuburblia/ Kotias (LUP/ Mesolithic sth caucasians) were not included as the 'Big 4', nor were Iran Neolithic (incl. the oldest one WC1), or Hotu Mesolithic - as they too can be modelled as a product of MA1, Natufians, Onge/ U-I.

In the table it can be seen that the major difference between CHG (KK/ Satsb) and Iran neolithic / Hotu is that the latter 2 posses Onge/ U-I (ie proto-ASI), whilst Kotias / Satsrb do not.
What catches the eye is that ANE admixture (in the form of MA1 or AFG) appears to be prevalent from south of the Caucasus to the Zagros already by the late Upper Palaeolithic. I wonder if existed in a kind of mutual reverse gradient with Onge-like groups from south India to central Asia ??? (see below #2)
(Mesolithic samples from central Asia, and soon India, are on the way..)

Visually, and adding Levantine Ancients, at this gross overview we can discern 3 major axes:
13489

#2) Now I remove Satsurblia from the "target Pops" and add it as a source. After all, it was a real, existing UP group, so limiting modelling to MA1, U-I, and Natufians is- whilst interesting-ultimately artificial.
Now, the MA-1/ Afantova component seen in LUP / early Holocene west Asians disappears, as it is swallowed up by Satsurblia.

13490

But, it is present in modern South Asians (significantly) and modern Iranians (mildly). When did it arrive ?


#3) Adding in Iran Neolithic WC1 as a source also (existed by 10000 BC), removing Iran Neolithics from target groups.

13491

Adding Iran Neolithic diminished the prevalence of Kotias, because it becomes subsumed within Iran Neolithic itself.
However, Iran Neolithic drops markedly by Bronze Age in Armenia, suggesting new arriving groups/ population turnover, and continued weak presence in modern Armenians. Here extra Kotias/ 'CHG' is required.
Moving to Iran, whilst Iran Neolithic is strongly present in Iran Chalcolithic, we also see the programme instead choosing extra Kotias, as well as a rise in overall Levant / farmer ancestry. We shall see why soon

#4) the "K4000BC". Further remove Iran Chalcolithic and Armenia Chalcolithic from target groups, and add it into source:

13492

Now we again see Iran Neolithic ancestry plummets to virtually zero in Armenian BA, and remains so in modern Armenians.
Instead, it is specifically Iran Chalcolithic which is the preferred ancestral component in Bronze Age Armenia, modern Armenians & modern Iranians.
Yet for India and the Afghano-Pakistan area, it is still Iran Neolithic which is relevant, not Iran Chalcolithic.
This means that northern India was somehow removed from the various events of local replacements and admixture which characterised chalcolithic northern West Asia.

Gravetto-Danubian
01-06-2017, 07:43 AM
#5) Adding 3000 BC samples to source Groups: Yamnaya/ Afansievo, EBA Armenia, Jordan EBA.

13493

- Iran Neolithic still pertinent for India/ Balochistan, etc
- Iran Chalcolithic diminishes at expense of Armenia EBA, but is still present in Armenians, Iran Manzand., the Kurd, and very significantly in the Iron Age Iranian.
- Yamnaya admixture now appears in place of Afantova or Samara HG in India (but choppy), and some Iranians (Mazand, Zoro) and 10% in the Iron Age Iranian.

#6) Throwing in the last lot: MBA Armenians, the Iron Age Iranian, Scythians, Srubnaya, Sintashta, Iron Age Altai, etc..2200 BC and beyond:

13495

- similar patterns with Iran Neol, Arm EBA.
- the Iron Age Iranian has best preserved ancestry in Armenians, whilst its input into modern Iranians seems negligible. on the PCA it was too "farmer" shifted, so perhaps represents and outlier, or we are in the dark about other LBA - Iron Age events in Iran.
- we see why modern Armenians are 'de-stepped' compared to MBA- mostly Anatolian-farmer -like admixture (? from where)
- impact of "Jordan BA" like population on Iranians, incl Iraqi Kurd.
- significant Scythian impact on modern Iranian (10-20%)
- the Kurd always looked similar to Armenians, apart from its 'exotic' IA Altai admixture (? historical Turkics) and Levant EBA shift instead of LBK/Barcin shift.
- NB: somthing add has happened to the steppe admixture representation. The programme chose Samara HG instead of Yamnaya, or Stubnaya or Sintashta or Andronovo. Perhaps because we don;t have the suitable choice yet.


_______
As a tentative conclusion, the "waves of invasion" / admixture in northern West Asia can be characterised as;

1) essentially Satsurblia -like population inhabiting the Iranian plateau, mixing with Onge-like south Asian foragers (10-20%). This was formulated by the Neolithic.

2) by the Chalcolithic, there is shifting/ admixture by a population characterized by higher 'pure' Kotias and Levant/ Anatol. farmer ancestry- noted in the Lazarides paper. As Davidski once suggested, this could be the ~ Halaf expansion. This displaced the older stratum of Iran Neolithic in Iran itself, but the original "Iran Neolithic" ancestry was very well preserved to this day in Balochistan, and northern India, Hindu Kush, as per the Mark Thomas et al paper. (This expansion, or similar, also affected Armenia and the emerging Anatolian Chalcolithic)

3) c. 3000 BC. The expansion of the Kura-Araxes system, impacting from the northern Levant to Iran and Pashtuns and Tajiks, but no impact on India itself.

4) Evidence of "steppe impact" in India, peaking in kalash, but attenuates in Iranian plateau. Given that it is only 10% in Iron Age Iran, perhaps half of the 10-20% steppe in modern Iranians came from historical Scythians. Steppe impact also notable in Armenia MBA, but choppy, suggesting selective admigration. It drops significantly by modern Armenians, due to yet to be described events, but points to west of Armenia.

5) Significant Jordan EBA impact in Iran, perhaps with historic movements and cultural events.

This is a 'screen' for more formal tests, direct DStats, IBD, etc. Of course, more samples would be great too.

Gravetto-Danubian
01-06-2017, 08:11 AM
Would you mind posting your nMonte population spreadsheet? I'd like to run my family and the other member samples I have from the global 10. Your population choices seem well thought out; curious to see how they compare to what you've already posted.

I will probably check back in a few days. Been busy as of late. Thanks for posting, or considering so, in advance.

What I use is in the Globe10, plus some extras. IN my above post to Kurd, I set out how I construct different source pops, on a period by period basis, and can be seen in the attached Xcel sheet.
If you still need clarification, just ask, or ask what you want tested, Ill email you a created source.txt file

Kurd
01-06-2017, 01:54 PM
@ Gravetto

Thanks for taking the time to do the nMonte runs. Your procedure and deductive reasoning is good, unlike many others here who seem totally clueless....

Based on the outputs I saw for Kurd Iraq, there appears to be limitations and issues with nMonte outputs, whether they are ADMIXTURE based or PCA. One of the limitations I had pointed out at http://http://www.anthrogenica.com/showthre...421#post201421

Another problem is that ADMIXTURE and PCA are themselves very volatile where the results vary based on the number and type of samples in the runs.

My recommendation is to treat any nMonte model with caution, and to verify the output with formal tests, such as dstats.

For example, nMonte showed Kurd Iraq had 0 Anatolia N, which seemed odd to me. Dstats showed that to be incorrect, because Kurd shared more alleles with Anatolia N than with Levant N and CHG (dstats on previous page). Also, nMonte shows Kurd Iraq with much less Iran N than the Iranian samples. Direct sample to sample comparisons with Dstats, which are very reliable BTW, showed the opposite. They showed Kurd shared about the same number of alleles with Iran N, as Zoroastrians and Fars samples share with Iran N, if not insignificantly more.

The same with Scythian IA being 0 (although nMonte showed 9% Altai for Kurd instead). The following dstats show that Kurd Iraq shared more alleles with Scythian IA than any of the W/SC/S Asian samples in the table.



W/SC/S Asians
Kurd Iraq
TARGET
OUTGROUP
D
Z
SNPs


Levant_N
Kurd Iraq
Scythian_IA
Chimp
-0.0488
-5.332
70060


Iran_N
Kurd Iraq
Scythian_IA
Chimp
-0.0428
-5.019
79368


Kostenki14
Kurd Iraq
Scythian_IA
Chimp
-0.0405
-4.72
91700


Jordanians
Kurd Iraq
Scythian_IA
Chimp
-0.0375
-6.038
98625


BedouinB
Kurd Iraq
Scythian_IA
Chimp
-0.0366
-6.859
98625


Iranian_Bandari
Kurd Iraq
Scythian_IA
Chimp
-0.0346
-6.715
98419


Iran_N_WC1
Kurd Iraq
Scythian_IA
Chimp
-0.0312
-3.654
61880


GujaratiD
Kurd Iraq
Scythian_IA
Chimp
-0.0273
-5.052
98625


Kotias
Kurd Iraq
Scythian_IA
Chimp
-0.0188
-2.283
98604


Druze
Kurd Iraq
Scythian_IA
Chimp
-0.0187
-3.263
98625


Iran_ChL
Kurd Iraq
Scythian_IA
Chimp
-0.0176
-2.679
92951


Balochi
Kurd Iraq
Scythian_IA
Chimp
-0.0174
-3.454
98625


Pashtun_Afghan
Kurd Iraq
Scythian_IA
Chimp
-0.0171
-3.226
94992


Jew_iraqi
Kurd Iraq
Scythian_IA
Chimp
-0.0165
-3.125
98625


Levant_BA
Kurd Iraq
Scythian_IA
Chimp
-0.0158
-2.191
83902


Iranian
Kurd Iraq
Scythian_IA
Chimp
-0.0136
-2.799
98625


Iran_Zoroastrian
Kurd Iraq
Scythian_IA
Chimp
-0.0125
-2.505
98554


Armenians
Kurd Iraq
Scythian_IA
Chimp
-0.012
-2.194
98625


Satsurbila
Kurd Iraq
Scythian_IA
Chimp
-0.0118
-1.31
69432


Pathan
Kurd Iraq
Scythian_IA
Chimp
-0.011
-2.209
98625


Georgians
Kurd Iraq
Scythian_IA
Chimp
-0.0099
-1.651
98625


Iran_recent
Kurd Iraq
Scythian_IA
Chimp
-0.0094
-1.05
75194


MA1
Kurd Iraq
Scythian_IA
Chimp
-0.0093
-0.965
71199


Turkish
Kurd Iraq
Scythian_IA
Chimp
-0.0087
-1.786
98625


Anatolia_N
Kurd Iraq
Scythian_IA
Chimp
-0.0084
-1.622
98625


Assyrians
Kurd Iraq
Scythian_IA
Chimp
-0.0076
-1.39
98625


Kalash
Kurd Iraq
Scythian_IA
Chimp
-0.0063
-1.153
98625


Azerbaijanis
Kurd Iraq
Scythian_IA
Chimp
-0.0061
-1.083
98625


North-Ossetians
Kurd Iraq
Scythian_IA
Chimp
-0.0028
-0.468
98625


Lezgins
Kurd Iraq
Scythian_IA
Chimp
-0.0023
-0.441
98625




nMonte shows Kurd Iraq with high Armenian EBA, which dstats do confirm, but is almost like if I got modeled with nMonte as 50% my father, but my goal was to see how much steppe ancestry, Iran N ancestry, etc. I had inherited through my father. In this case I would do direct comparisons using dstats between myself and steppe, Iran N, etc, to get an idea how many alleles I share with Iran N and steppe vs what other pops share with steppe and Iran N.

I will follow everything up later with IBD.

Iran Chl seem to have diluted some Iran N ancestry in W Iranians and Kurds, but their impact was negligible on SE Iran and eastward.

A curious factoid about Yamanya.... about 1.5 years ago, I had posted here that I believed Yamnaya was more important than Sintashta or Andonovo for some S/SC Asians, based on what I was seeing. This was long before David and others posted about the connection between Yamnaya and modern S/SC Asians. At the time, I was ridiculed because the Yamanya samples hitherto were R1b, and I kept pointing out that we only had a handful of samples covering a huge geographic area, and that other Yamnaya Y-lines will be forthcoming in the future....I feel a little vindicated now....

Gravetto-Danubian
01-07-2017, 06:03 AM
Absolutely. Triple checking with all methods is the way to go, especially if looking at specific groups/ individuals affinities. I think nMonte is reasonable for broad population trends, and has been more or less correct for all modelled populations (i.e. match academic papers).

Awale
01-07-2017, 06:22 PM
This is a serious charge that I don't take lightly. I hope you realize that I can easily bring a defamation of character law suit against you if I choose to. I have your post as evidence of the defamatory statement and the false statement of fact, to wit, "my results are doctored", and all I have to do is give my dataset to someone who knows and is able to run dstats, and show that they come up with the same results as me. As far as unveiling your identity and address, all I would have to do is subpoena this website's records from their domain host on the East Coast, and take my word for it I have alot of experience in this area.

In fact, you could be exposing this website to liability, since you have published a serious charge, without offering proof that my results are doctored.

I'll be replying to you strictly as a moderator here:

Firstly, who is Pegasus defaming? "Kurd"? Who's Kurd? Some Y-DNA T and mtDNA W guy of supposedly Iraqi-Kurd+Punjabi descent who lives god only knows where? You're speaking as though he's defamed you on live television with your true identity known to everyone. Hell, we can't even be sure these avatars you often share are really you or that you're even male or that your Y-DNA and mtDNA are yours or that you're even ethnically what you are. Secondly, the only people who could reliably trace Pegasus' identity for you are entities like his service provider and email source, not us or our domain host. You're completely welcome to tackle those sorts behemoths for that information though, by all means.

Thirdly, you should, in the future, save yourself the embarrassment and refrain from making anymore bogus and frivolous legal threats to silence opposition. If you feel like someone has overstepped the boundaries of either the law, or our own ToS, report them and contact the staff directly (PMs) for further questions. Finally, I, and the rest of the staff, don't appreciate you possibly spooking anyone on this forum with threats of any sort and let that be known to you, my friend. Any more of such public threats will result in membership sanctions.

Update:

It should be noted that Pegasus has caused trouble here as well. What he did by "calling you out" (his own words) unnecessarily personalizes the discussion which is against our terms of service:


3.12 Anthrogenica encourages its members to participate in discussions in a topic-focused manner. Personalization of discussions is completely prohibited at all times. This includes (and is not limited to) direct personal attacks, accusations, insinuations and false disclosures. Additionally, discussions that degenerate into inconsequential flaming or inanity will be deleted without prior notice.

Moderator
01-08-2017, 01:36 AM
[MOD]

Several more OT posts deleted. A member of staff has already weighed in above on the needlessness for this tangent. Any more posts of this nature will be deleted without warning and infractions will follow.

As a reminder - ALL members are expected to uphold the Terms of Service through their continued participation in this forum. In continuation with Moderator Awale's input, registered users must adhere to the Terms in their entirety. In this instance:



3.12 Anthrogenica encourages its members to participate in discussions in a topic-focused manner. Personalization of discussions is completely prohibited at all times. This includes (and is not limited to) direct personal attacks, accusations, insinuations and false disclosures. Additionally, discussions that degenerate into inconsequential flaming or inanity will be deleted without prior notice.


As a second reminder, please report problem content rather than taking matters into your own hands.

Thanks for your cooperation.

parasar
01-08-2017, 05:01 AM
... At the time, I was ridiculed because the Yamanya samples hitherto were R1b, and I kept pointing out that we only had a handful of samples covering a huge geographic area, and that other Yamnaya Y-lines will be forthcoming in the future....

Even if they don't come out as R1a it does not make a difference. R1b is nearly as close to R2 as it to R1a. It is quite possible that R1b is older than R1a in South Asia and that the earlier (pre-steppe intrusion) population of South Asia was quite Yamna like.

Kurd
01-09-2017, 07:13 PM
Who's Kurd? Some Y-DNA T and mtDNA W :

Please remove my privacy protected information from your post. I have not given permission to you or anyone else to post. It is irrelevant if I had included any such info with my profile at some point. If I wanted to share this, I would include it in my current profile. Also, not only does the discussion about my address or ethnicity absolutely bear no relevancy to the issue at hand, it is not appropriate to insinuate what information may be false, especially coming from a person who is held to a higher standard, such as a moderator.

No need to remove this post, since nothing herein violates the site's TOS.