04-27-2024, 03:55 AM (This post was last modified: 04-27-2024, 04:02 AM by Jerome.)
(04-26-2024, 05:21 PM)TanTin Wrote:
(04-26-2024, 12:04 PM)Jerome Wrote: Tantin,in the first image satsurblia.sg has a Z score of 0.990 while kotias is reaching 2.11
Both are lesser than turkey N and slightly lesser than iran_n too.
Does this mean Satsurblia has the least Neanderthal here and most basal?
Also,could you test this same model using Wezmeh,Abdul hosein and hotu?
Tepe guran would be interesting but I guess allentoft isn't merged yet
There was a bug with the converted dataset, that I reported in another topic. So I am not sure if these results were from that set with the bug. I have to re-do the test to verify if these are the correct numbers.
Here it does seem to be different and not as low Z scores that you were getting for satusrblian before(0.8).
Satsurblia is getting 3.1 Z score and Kotias 5.43 Z score now.
But satsurblia is still pretty low relatively to others and even lower than Israel_PPNB/Levant_N and Iberomaurusians?
Is this an artifact of sampling method of satsurblia or is it real?
Could you also redo this and add Wezmeh_N.SG,Hotu and Tepe Abdul hosein samples?
I am trying to see which iranN has highest Neanderthal affinity.
Pinarbasi_Epipaleolithic,Kotias_UP/Dzudzuana and Tepe Guran too maybe,if you have it.
(04-16-2024, 04:54 AM)Jerome Wrote: If CF is crown Eurasian then DE must be basal eurasian and their ancestors (proto-eurasians) should be CT.
The confusion regarding ANA and taforalt can be cleared with the fact that a)ANA wasn't African proper and instead seems to be an early split from the proto Eurasian group ,it should also be closer to proto eurasians than to east Africans like mota
B)ANA likely had mixed with back-migrating E carrying basal eurasians before the iberomaurusian profile formed.
So the E in IBM could be from the basal part,for all we know.
We need more samples to discern these things but still it's a plausible theory.
Considering that CF and D are almost fully Eurasian ,I find it too much of a stretching to say that CF and D left africa leaving only E rather than simply say CF and DE were born somewhere on the border of Eurasia and Africa (probably sinai,NW Saudi arabia) and then E stayed with the basal Eurasian group and later mixed with Africans while D went alongwith crown Eurasian groups.
The original African groups would be subclades of A0,A,BT and B and got E after basal Eurasian introgression around 50k BC(there was a study which mentioned that the minor Neanderthal alleles in modern Africans date to this period 50-40k BC).
This is the study http://dx.doi.org/10.1016/j.cell.2020.01.012
This theory is supported by the fact that the most archaic Africans with early splits seem to be A and B like ju hoang and biaka while those near Eurasia/Nile Valley seem to be E (east Africans).
E in west Africans seems to be very recent with most TMRCA dating after LGM and mostly afterr Holocene.
Instead of searching for basal Eurasian in the Persian gulf which was surrounded by Neanderthals(shanidar,zarzi,warsawi,wezmeh) and where there are barely any sites in MIS-5/MIS-4 ,we should instead look for basal Eurasian on the border/periphery of Africa and Asia,that is,somewhere in Sinai,Nile valley,Libyan coast or NW Saudi arabia/upper hijaz where the basal split likely happened (if we consider that the emiran people of levant were crown Eurasian like,(emiran->bohunician°>bacho kiro ) ,then we would expect basal eurasians to be a bit westwards from the levant and probably in Nile valley,Libyan coast.
E in natufians could be from the basal Eurasian+ANA mixed Taforalt group or b)could be from the original basal Eurasian admixture in kebarans.
We need more samples to clarify this, especially from paleolithic levant/kebarans.
Epi paleolithic north levantines should work for this job too.
(04-24-2024, 12:06 PM)Desdonas Wrote: Latitude may truely distinguish the Ust-Goyet lineage and East Eurasian lineage. But I think that their common ancestor (Crown/C-K meta population) arose in eastern Iran Plateau. Anyway, AASI and Australasians should all head south, since they don't have the northern Denisova admixture (D0).
Interesting,does this mean AASI and southern east eurasians didn't take the central asia->altai->amur->south path?
Most people had been thinking this.
Out of curiosity; is the conclusion of AASI lacking D0 admixture based on AASI-rich moderns like Paniyas?
(04-27-2024, 03:55 AM)Jerome Wrote: Here it does seem to be different and not as low Z scores that you were getting for satusrblian before(0.8).
Satsurblia is getting 3.1 Z score and Kotias 5.43 Z score now.
But satsurblia is still pretty low relatively to others and even lower than Israel_PPNB/Levant_N and Iberomaurusians?
Is this an artifact of sampling method of satsurblia or is it real?
Could you also redo this and add Wezmeh_N.SG,Hotu and Tepe Abdul hosein samples?
I am trying to see which iranN has highest Neanderthal affinity.
Pinarbasi_Epipaleolithic,Kotias_UP/Dzudzuana and Tepe Guran too maybe,if you have it.
To understand better for Neanderthal / Denisovan I prefer to use another F4 stats.
f4( prefix, "Denisova.DG" , "Vindija_Neanderthal.DG" , "Yoruba.DG" , TEST) #
This one will measure the distance to Denisova / Neanderthal and will sort all the samples according to the position of Yoruba. Yoruba is our 0-line. Those who are similar to Denisova will go on one side, Neanderthals related will go to the other side.
This one is more accurate compared to f4( prefix, "Chimp" , "Vindija_Neanderthal.DG" , "Yoruba.DG" , TEST) # We may use Chimp for other tests and verifications, depending on our goals.
So let me show you what do we have for Denisova / Neanderthal check:
To visualize I will sort them from South Africa to Asia / Japan.
Here I visualize them all, including Denisova, Chimp and few Neanderthals. As you may see: we don't see much details, because Neanderthal component is very low. But we see that almost all are on the side of Neanderthals. Except Chimp, who is archaic and who has similar archaic snips as the Denisovan. Also we know that Denisova is closer to the Chimp. Neanderthals are more distant from the Chimp.
Now I am going to remove the first 5, so that will allow us to see more details for the Neanderthal components in the above populations.
BTW, these results are generated from V52.2 - v52.2_1240K_public dataset.
You may also see that more than 1 000 000 snips were used.
04-27-2024, 09:52 PM (This post was last modified: 04-27-2024, 10:22 PM by TanTin.)
Pay attention, that the most Neanderthal individuals are Bacho Kiro.
If I add Oase, it will show even more Neanderthal, but Oase is not included in v52.2_1240K . It is in the other datset for HO.
The most Denisovan should be Papuans, but somehow in my results I got Mbuti and Ethiopean as more Denisovan than Papua.
(04-27-2024, 03:55 AM)Jerome Wrote: Here it does seem to be different and not as low Z scores that you were getting for satusrblian before(0.8).
Satsurblia is getting 3.1 Z score and Kotias 5.43 Z score now.
But satsurblia is still pretty low relatively to others and even lower than Israel_PPNB/Levant_N and Iberomaurusians?
Is this an artifact of sampling method of satsurblia or is it real?
Could you also redo this and add Wezmeh_N.SG,Hotu and Tepe Abdul hosein samples?
I am trying to see which iranN has highest Neanderthal affinity.
Pinarbasi_Epipaleolithic,Kotias_UP/Dzudzuana and Tepe Guran too maybe,if you have it.
To understand better for Neanderthal / Denisovan I prefer to use another F4 stats.
f4( prefix, "Denisova.DG" , "Vindija_Neanderthal.DG" , "Yoruba.DG" , TEST) #
This one will measure the distance to Denisova / Neanderthal and will sort all the samples according to the position of Yoruba. Yoruba is our 0-line. Those who are similar to Denisova will go on one side, Neanderthals related will go to the other side.
This one is more accurate compared to f4( prefix, "Chimp" , "Vindija_Neanderthal.DG" , "Yoruba.DG" , TEST) # We may use Chimp for other tests and verifications, depending on our goals.
So let me show you what do we have for Denisova / Neanderthal check:
To visualize I will sort them from South Africa to Asia / Japan.
Here I visualize them all, including Denisova, Chimp and few Neanderthals. As you may see: we don't see much details, because Neanderthal component is very low. But we see that almost all are on the side of Neanderthals. Except Chimp, who is archaic and who has similar archaic snips as the Denisovan. Also we know that Denisova is closer to the Chimp. Neanderthals are more distant from the Chimp.
Wouldn't this be a bit flawed for ENA rich pops like iranN and CHG?
The denisovan in these populations would deviate the results because of the outgroip.
I would say chimp would be better for these populations.
(04-28-2024, 05:08 AM)Jerome Wrote: Wouldn't this be a bit flawed for ENA rich pops like iranN and CHG?
The denisovan in these populations would deviate the results because of the outgroip.
I would say chimp would be better for these populations.
I agree with your comment.
Denisova components may flaw the results for some Asian groups. So let see what is the difference if we use Chimp instead of Denisova.
As we may see, there is no big difference, except for Papua. There is a small difference in MA1 as well, so we see there is some Denisova component in MA1. Others don't seems to be much affected . Iranians don't have much Denisova and their Neanderthal component is about the average compared to other European and Asian groups.
To validate our results and to be sure in the previous test I run another fstat:
"F4 ( Chimp.REF Denisova.DG ; Yoruba.DG TEST )"
This will show us the amount of Chimp/Denisova components .
Chimp is very early and archaic. Denisova is much later in the evolution .
Early humans are branching out somewhere between Chimp and Denisova, but more on the side of Denisova. Other research may tell that Denisova are branching out before the humans. But that's not the goal of this Fstat. We are not going to redo and fix the tree at the moment.
We will directly test if our populations are more related to Chimp or to Denisova.
Here are the results. Dataset used: full dataset: v52.2_1240K
04-28-2024, 01:29 PM (This post was last modified: 04-28-2024, 01:53 PM by Jerome.)
(04-28-2024, 11:03 AM)Norfern-Ostrobothnian Wrote: I don't think CHG or Iran Neolithic have any East Eurasian beyond Ancient North Eurasian ancestry.
It's pretty complex and dubious, TBH.
Using zlatyKun as a source iran_n dosent need any ENA but using Pinarbasi,WHG or Dzudzuana does.
Here's a model by the mnemonics guy from anthrogenica.
"MNEMONICS Wrote:I wonder if the "Onge-like" affinity in Iran_N that distinguishes it from CHG is actually real, I don't have much trouble modeling Iran_N as CHG + ZlatyKun + Tyumen + Admixed NE African with Indian_GreatAndaman_100BP in the right popslist. Assuming that the ZlatyKun + NE African is acting as a proxy for Basal, there doesn't seem to be much need for extra Onge related ancestry.
right pops:
Eastern_Euro_UP.SG
Cameroon_SMA.SG
South_Africa_400BP.SG
Mota
South_Africa_HG
Indian_GreatAndaman_100BP.SG
DevilsGate_N.SG
Yana_UP
BachoKiro
Nepal_Samdzong_1500BP.SG
Kazakhstan_Botai_Eneolithic.SG
Boncuklu_N.SG
Morocco_EN.SG
MA1
Italy_Mesolithic.SG
P-value 0.257687 Best coefficients 0.630 0.258 0.055 0.057
63% CHG,26% Zlaty,5% East Africa pastoral,6% extra ANE
...
I think the issue is actually very different instead,from checking the Neanderthal levels and affinity ,iran_n dosent seem to have exceptionally low Neanderthal Ancestry or high basal Ancestry (Lazaridis in his dzuduana preprint modelled iran_n as 55% Dzuduana+ 10% extra mbuti/basal,and Dzuduana being 26% Basal in his study,this 55% Dzudzuana+10% Mbuti translates into 24% Basal for Iran_N,22% for CHG).
This is radically different from his earlier estimates in 2016 where he postulated iran_n as being '48% Basal eurasian'.
It seems what changed here was accounting for the ENA admixture in iran_n which he didn't do before and the ENA like affinity was simply getting transferred to Basal Eurasian,since Lazaridis didn't add an ENA source in his 2016 study and used the the F4 statistic (Chimp UST Ishim X WHG/Kostenki).
This form has many phylogenetic constraints especially if iran_n derives from a branch that split after ust ishim but earlier than WHG/Kostenki.
Lazaridis actually pointed this out but went alongwith his dubious conclusions anyways.
This led me to think more of this possibility.
Here's a lis of posts from Kale and K33 on this concept who explored it a while before for a short period.
KALE Wrote:- Iran_N & CHG derive most of their ancestry from a branch (NE0/NE1)that combines the earliest West Eurasian split and some East Eurasian, with lesser amounts of ANE, and a WHG-cousin on top of that. - Barcin_N is less NE0, way more WHG-cousin, no excess ANE, a bit of Taforalt. - EHG = ANE + WHG + a bit of NE0/NE1
KALE Wrote:
K33 Wrote:The only ancient Near Eastern population modeled successfully here without Basal/para-Eurasian ancestry is Iran_N,
And CHG too.
When all mixtures are considered, Iran_N comes out 71.74% NE1. NE1 is the earliest split in West-Eurasians, and contains a bit of East-Eurasian ancestry as well.
ANE was on the border of East and West in the North, so that leaves NE1 inhabiting the East-West border on the South, which I would presume to be South-Central Asia.
That looks like something analogous to your ACE concept.
There definitely is backflow into Iran_N from ANE though, y-hg R2 being evidence of that.
From this thread-https://genoplot.com/discussions/topic/29484/survival-of-late-pleistocene-hunter-gatherer-ancestry-in-the-iberian-peninsula/30
K33 Wrote:To put it another way: East and West Eurasians split no later than 45kya, and then ANE split from West Eurasian no later than 32kya (Yana RHS). Isn't it more likely Iran_N and CHG derive mostly instead from a unified Central Eurasian HG population ancestral to both ANE and Sunghir/Kostenki? Such "Basal Crown Eurasian" ancestry would also help explain the falsely elevated signals of BE ancestry (originally pegged by Lazaridis at 50%) in Iran_N. Though I still think Iran_N has SOME genuine BE ancestry-- just significantly less than Barcin and Natufians
K33 Wrote:
KALE Wrote:But Ganj_Dareh_N (Iran_N) derives 0% from NodeA in that graph. Even if you wanted to call NodeA basal, Iran_N still need no basal there.
True, but Ganj_Dareh DOES take 52% ancestry from "West1", which is a mere 55 drift units away from NodeA. 52% of Gan_Dareh's ancestry traces directly to a source only 198 drift units (143+40+15) from NodeA. And 66% of Ganj_Dareh is derived from sources only 297 drift units from "Eurasia" (if you include the "East0" anchor directly contributing to NE1
..
Kale Wrote:
K33 Wrote:True, but Ganj_Dareh DOES take 52% ancestry from “West1”, which is a mere 55 drift units away from NodeA.
A true ‘basal Eurasian’ should be equally related to all ‘crown Eurasians’, but less related to crown Eurasians than crown Eurasians are to each other. West1 does not match that criteria. It shares 40 units of drift with archetypal West Eurasian populations.
If West1 were to be sampled in pure form, it would be more related to say Kostenki14 than it would be to
Tianyuan.
Look at the other historical anchors:
GoyetQ116, Vestonice16, Kostenki14, Tianyuan all get 100% of their ancestry from “Eurasia” too-- but they all require 450+ drift units from “Eurasia” to account for their position.
An important technical detail of qpgraph is that any drift edge leading to a single sample is not informative. Compare those to the drift edge to Sunghir (multiple samples combined), the drift is only 42.
It seems to be an interesting concept combined with Haplogroup G in iran_n(First split in GHIJK,Formed 47K BC),I don't know why Kale stopped exploring this likely ancestry in iran_n.
I would wait for samples from badarostian culture/UP iran before postulating too much about basal Eurasian and ENA Ancestry in iran_n, and combined with the presence of Haplogroup G in iran_n,there does seem to be quite an evidence and probability of somekind of 'Basal West Eurasian/Early West Eurasian' in iran_n.
It could also explain the slight ENA affinities considering that it only appears when you use later differentiated west eurasians like WHG/Dzuduana/Pinarbasi to model iran_n and this 'ENA' like Affinity and also the excess Basal affinity may simply be pointing to an undifferentiated early West Eurasian form that didn't undergo the later drifting and transformation into the archetypical west eurasians symbolised by Sunghir and WHG.
We all know how Admixture calculators show Onge and AASI in some aurignacian samples and also in ust ishim even though phylogenetically there's none,this might be a similar case with the 'Basal+Mbuti' admixture in iran_n simply be a result of not having proximal sources.
04-28-2024, 01:50 PM (This post was last modified: 04-28-2024, 01:55 PM by Jerome.)
Thinking of this,I think it would be interesting to see how IRAN_n models on QpAdm if Kostenki14+Zlaty Kun/Ust ishim+ANE/MA1 are used in left pops alongwith Dinka and Onge added in left pops to see if there's still any contribution from Dinka(Deep/Basal proxy) and Onge if Zlaty/Ust and Kostenki are used in left pops to account for the archaic west Eurasian affinities that can't be provided by dzudzuana/WHG.
Best outgroup for this model would Be SA_2000BP and the other usual list of right pops.
04-28-2024, 03:46 PM (This post was last modified: 04-28-2024, 04:01 PM by Kale.)
(04-28-2024, 01:29 PM)Jerome Wrote: It seems to be an interesting concept combined with Haplogroup G in iran_n(First split in GHIJK,Formed 47K BC),I don't know why Kale stopped exploring this likely ancestry in iran_n.
That was a long time ago. ZlatyKun was the main reason I stopped that line of inquiry. Up until that sample, I didn't see a need for Basal Eurasian. Not sure how ZlatyKun would fit in a graph where Iran_N has no Basal. I'll give it some thought on the road today.
(04-28-2024, 01:50 PM)Jerome Wrote: Thinking of this,I think it would be interesting to see how IRAN_n models on QpAdm if Kostenki14+Zlaty Kun/Ust ishim+ANE/MA1 are used in left pops alongwith Dinka and Onge added in left pops to see if there's still any contribution from Dinka(Deep/Basal proxy) and Onge if Zlaty/Ust and Kostenki are used in left pops to account for the archaic west Eurasian affinities that can't be provided by dzudzuana/WHG.
Best outgroup for this model would Be SA_2000BP and the other usual list of right pops.
Iran_Wezmeh_N.SG
ZlatyKun.SG 0.325104 0.0881079 3.68984
Kostenki14.SG 0.275399 0.0391516 7.03417
AG3 0.139199 0.0215189 6.46866
Andaman_100BP.SG 0.189066 0.0520910 3.62953
Mota.SG 0.0712322 0.0176885 4.02704
Tail: 0.13
right = c('SouthAfrica_2000BP.SG', 'Ust_Ishim.DG', 'BachoKiro_IUP', 'Papuan.DG', 'China_UP', 'RUS_Primorsky_DevilsCave_N.SG', 'Muierii1', 'BachoKiro_BK1653', 'Yana_UP.SG', 'MA1.SG', 'Peru_RioUncallane_1800BP.SG')
allsnps=TRUE
Goyet, Fournol, or Sunghir in right kill the model.
(04-28-2024, 01:50 PM)Jerome Wrote: Thinking of this,I think it would be interesting to see how IRAN_n models on QpAdm if Kostenki14+Zlaty Kun/Ust ishim+ANE/MA1 are used in left pops alongwith Dinka and Onge added in left pops to see if there's still any contribution from Dinka(Deep/Basal proxy) and Onge if Zlaty/Ust and Kostenki are used in left pops to account for the archaic west Eurasian affinities that can't be provided by dzudzuana/WHG.
Best outgroup for this model would Be SA_2000BP and the other usual list of right pops.
This is a good question and I have something to offer you:
"F4 ( Chimp.REF ONG.SG ; Yoruba.DG TEST )"
This will tell us how much Onge components are shared with the populations.
Using Chimp as Outgroup and Yoruba as base line (0-line).
Instead, here is fstat by using Iran Mesolithic . We may see that later Iranian groups contain ancestry from Iranian Mesolithic.
But there is not much if it in other groups outside of Iran.
![[Image: FsLwdRcXwAER5RC?format=jpg&name=large]](https://pbs.twimg.com/media/FsLwdRcXwAER5RC?format=jpg&name=large)
