search this blog

Wednesday, August 19, 2020

Yamnaya-related ancestry proportions in present-day Poles


Modeling ancient ancestry proportions in present-day Europeans with the qpAdm software is now a lot more difficult. The reasons for this are updates to qpAdm as well as the availabiity of more useuful outgroups or right pops.

This isn't necessarily a bad thing, because users are forced to work harder to find successful models, which is likely to lead to some interesting discoveries. But it can be very frustrating.

I don't think that settling for poor statistical fits or using a small number of outrgoups are acceptable short cuts. Perhaps sequencing modern-day samples in exactly the same way as the ancient samples, and thus increasing the compatability between them, might help?

Limiting qpAdm runs to higher quality SNPs from transversion sites does help, but perhaps largely because of the significant reduction in markers?

In any case, I've now given up on running such analyses, at least until I see some serious pointers on the topic from Harvard's qpAdm experts. But before I put this project to bed for the time being, I'd like to share some new results for Poles from eastern and western Poland, respectively.

right pops:

CMR_Shum_Laka_8000BP
MAR_Taforalt
IRN_Ganj_Dareh_N
Levant_PPNB
GEO_CHG
TUR_Barcin_N
RUS_Piedmont_En
SRB_Iron_Gates_HG
WHG
RUS_Karelia_HG
MNG_North_N
RUS_Ust_Kyakhta

left pops:

Polish_East
CWC_Baltic_early 0.572±0.024
SWE_TRB 0.428±0.024
chisq 11.776
tail prob 0.300296
Full output

Polish_West
CWC_Baltic_early 0.587±0.021
SWE_TRB 0.413±0.021
chisq 11.165
tail prob 0.34478
Full output


Even using transversion sites, this is one of the very few combinations of ancient reference samples that works for the Poles with these right pops. That is, the combination of early Corded Ware samples from the East Baltic (CWC_Baltic_early) and Funnel Beaker samples from Scandinavia (SWE_TRB). The former are obviously the proxy here for Yamnaya-related ancestry.

Adding any sort of hunter-gatherer population to this model doesn't help or even makes things worse (for instance, see here and here). It is possible to add Baltic hunter-gatherers to a similar model after dropping CWC_Baltic_early in favor of closely related samples from the Early to Middle Bronze Age Pontic-Caspian steppe. Note, however, that the statistical fits are somewhat poorer.

Polish_East
Baltic_LTU_Narva 0.032±0.014
PC_steppe_EMBA 0.483±0.019
SWE_TRB 0.485±0.019
chisq 17.143
tail prob 0.0465198
Full output

Polish_West
Baltic_LTU_Narva 0.031±0.011
PC_steppe_EMBA 0.491±0.015
SWE_TRB 0.477±0.016
chisq 22.444
tail prob 0.00757421
Full output


Interestingly, but not surprisingly, the ancestry of many present-day Northwestern European populations can be modeled in basically the same way. That's because ancient ancestry proportions are more closely correlated with latitude than longitude across much of the European continent.

English_Kent
CWC_Baltic_early 0.527±0.024
SWE_TRB 0.473±0.024
chisq 13.042
tail prob 0.221357
Full output

Icelandic
CWC_Baltic_early 0.586±0.023
SWE_TRB 0.414±0.023
chisq 16.517
tail prob 0.085751
Full output

Scottish
CWC_Baltic_early 0.583±0.021
SWE_TRB 0.417±0.021
chisq 12.144
tail prob 0.275536
Full output


A zip file with the qpAdm output from this analysis and a list of the most relevant ancients is available here. I might try to run a few more populations over the next few days, but probably only from the northern half of Europe, so please check the zip file in a week or so to see what else is in there.

If anyone wants to challenge my results, note that these and very similar samples are freely available to the public via Harvard University here and here.

Update 22/08/2020: From Nick Patterson (Broad) in the comments:
My general advice for qpAdm is 1) Work on the right hand set. Don't include irrelevant population (except for one population as an outgroup); picking the best RHS can dramatically reduce s. errors on the admixture weights. 2) If qpAdm gives a very low p-value try and understand why, sometimes it is telling you that the target is not a mixture of the sources but sometimes the assumptions are violated, for example recent gene-flow from left pops -> right.

See also...

Ancient ancestry proportions in present-day Europeans

106 comments:

John Thomas said...

What does this all mean ?

At root, are all western Europeans really just transplanted central and eastern Europeans?

John Thomas said...

David, would it be possible to use these same basic, basic outgroups - together with appropriate 'local' basic outgroups, to gauge just how much 'Euro origin ancestry' exists in extant Indians, Iranians, Turks, Kurds etc.

andrew said...

So roughly equal parts early Bronze Age Ukrainian and Scandinavian herder-fisher, with a smidgeon (maybe 5-10%) of the indigenous Northeastern European local color that distinguished Corded Ware from Yamanya, right?

vAsiSTha said...

Nice set of outgroups, sturdy. But whats the difference between Srb_Iron_Gates_HG and WHG?

If scots, icelanders and poles have the same model and exactly same proportions, what explains the language difference if one goes by the 'genes dictate language' theory that you propose?

Davidski said...

@vAsiSTha

Iron Gates has more eastern EHG-like ancestry than WHG.

And of course there are genetic differences between Poles and Icelanders, Scots etc. But they're fine scale differences largely based on genetic drift that happened after the Corded Ware culture broke up.

Samuel Andrews said...

@John Thomas,
"At root, are all western Europeans really just transplanted central and eastern Europeans?"

The secret is Corded Ware. Which is what Davidski was telling you in a few sentences.

This makes sense when you put things in perspective. Corded Ware's range spread from Netherlands to Moscow. So it makes sense they are the shared link between East and NorthWest Europeans. Corded Ware ancestry is not unique to them though, every European ethnic group has significant Corded Ware ancestry.

Corded Ware is source of most European languages.

The old idea, was many different migrations from Steppe spread Indo European languages in Europe. Ancient DNA shows is it was mainly Corded Ware. The old IE languages of Southeast Europe, most importantly Greek, don't come from Corded Ware. But, all others do including Latin.

This makes sense, because Corded Ware were Indo European people who migrated/invaded the heart of European continent and spread across the whole Northern half of it. They covered a large area. They didn't live side by side native populations. They conquered, assilimated, pushed out the native populations in the process. It was a major demographic event, not just a change in pottery style. It was an Indo European take over of half of Europe. Then Bell Beaker, continued this invasion into Western Europe.

Corded Ware mixed with farmer populations across Central/East/Northwest Europe who known as Globular Amphora & Funnel Beaker. GAC and TRB were essentially the same people (GAC came from TRB). Because, of this not only do modern Northwest and CentralEast Europeans share the same Steppe ancestor but also some of the same farmer ancestors.

But there are important local differences in farmer ancestry. Clearly Northwest Europeans have significant Northwestern/Western Farmer ancestry which is somewhat distinct. And clearly Northeast Europeans have significant local hunter gatherer ancestry.

For most part Balto-Slavs are a product of Eastern Corded Ware, Northwest Europeans are product of Northwest Corded Ware. All Europeans north of Mediterranean coast are roughly similar products of Corded Ware and the various older pops they mixed with across Europe.

Slumbery said...

I just run this scenario with G25 nMontes. Not to challenge anything, but if anybody interested in the differences. Unlike the recent Barcin + WHG + EHG + Yamnaya run, this time there is no significant difference between nMontes and qpADM results.

"sample": "Polish:Average",
"distance": 5.5434,
"Corded_Ware_Baltic_early": 58.5,
"SWE_TRB": 41.5

"sample": "Icelandic:Average",
"distance": 3.5853,
"Corded_Ware_Baltic_early": 56.5,
"SWE_TRB": 43.5

"sample": "Scottish:Average",
"distance": 3.3172,
"Corded_Ware_Baltic_early": 53.5,
"SWE_TRB": 46.5

Reaction to Narva is very similar too.

Slumbery said...

Note however that in G25 nMontes if I add a combination of Carpathian Basic farmers + Narva to the mix, it eliminates Swedish TRB for Poles, but not for NW Europeans.


"sample": "Polish:Average",
"distance": 4.3947,
"Corded_Ware_Baltic_early": 52,
"SWE_TRB": 0,
"Baltic_LTU_Narva": 15,
"Lengyel_LN": 33

"sample": "Icelandic:Average",
"distance": 3.5014,
"Corded_Ware_Baltic_early": 55,
"SWE_TRB": 42,
"Baltic_LTU_Narva": 3,
"Lengyel_LN": 0

"sample": "Scottish:Average",
"distance": 3.3107,
"Corded_Ware_Baltic_early": 54,
"SWE_TRB": 43.5,
"Baltic_LTU_Narva": 0,
"Lengyel_LN": 2.5

So I do not think this farmer ancestry of modern CWC derived populations is really fully uniform, but TRB is good for a two way model.

Angantyr said...

@andrew

"Scandinavian herder-fisher"

TRB were farmers, not herder-fishers.

@Samuel Andrews

"They didn't live side by side native populations."

Actually, they did, at least in the north. It wasn't until post-CWC times that the pre-existing coastal/wetland HG populations disappeared from places like Scandinavia and the East Baltic.

Matt said...

Related, post inspired me to try something I haven't done before, see what European population structure looks like from a f3 outgroup perspective using the "MDS on f3" method used by some papers on structure within ancient HG.

Method: Take f3 matrix of modern-European populations, then use PAST3's principal components (MDS) function, taking the f3 matrix as a matrix of similarities. Graphically: https://imgur.com/a/nTEWyvD

Results 1: https://imgur.com/a/pfmzZML

That's strongly dominated in first dimensions by some populations with strong Siberian ancestry (Saami, Udmurd, Mari) and by some populations with North African / Near Eastern ancestry (Sicily and Jewish). These methods would be particularly sensitive to *any* geneflow which has any relatedness to the outgroup.

Removing those 6 groups: https://imgur.com/a/10x5RQk

So the interesting bit to me is that the structure of the Northern European groups (West: Irish, Scottish, and East: Ukrainian, Belarusian) with respect to their position in these plots relative to the Southern European groups (Spanish, Tuscan, Bergamo, Albanian) still persists, even with these formal stats. E.g. they're not positioned as if they were a clade that then drifted apart, with respect to the S European groups.

(f3 pastebin, past3 format: https://pastebin.com/6jKzjtKn)

Davidski said...

Apparently, f3 stats do pick up recent drift. It's the f4 stats and especially D stats that go deeper into the phylogeny.

But I can't recall the details now.

Davidski said...

@All

I've updated the zip file with the qpAdm output.

Latvians and Lithuanians do show significant extra hunter-gatherer ancestry in this model, which shouldn't be surprising, but they also show very high levels of early Corded Ware ancestry.

Uralic populations produced rather bad statistical fits in this model, even after adding Baikal_BA or Bolshoy_Oleni_Ostrov.

There's probably something important missing for them, like a Nganasan-related population that hasn't been published yet.

Chevalier de Balibari said...

I had a conversation with Onur Dincer about vikings lately.David do you think Vikigs had some Finnish-Estonian like admixture?And ofc Finns have some Germanic like input right?

I was checking this blog btw.

http://terheninenmaa.blogspot.com/2020/08/popular-vahaduo-oracle-for-finns.html

I think Vikings were raiding with Finns and Balts together.The theory that they were 100% germanics/norsemen is a little bit wrong i guess.There were some Uralic speaking among them and probably balts i would say.Any thoughts?

Davidski said...

Scandinavians do have some Uralic ancestry, both from Finns and Saami. There are different reasons for this.

Balts didn't have very much to do with Scandinavians, but Poles and Polabian Slavs did.

There were some very strong contacts between the Danish and Polish kingdoms, and Danes probably mixed quite a bit with the Slavs who lived along the Baltic coast and on nearby islands.

Samuel Andrews said...

@Angantyr,

There are exceptions to the rule. But, the rule is Corded Ware represents Indo European conquest of half of Europe. I'm open to any shift to this view. I would like see archeaologist combine ancient DNA and archeaology to understand what happened.

What I know, conquest makes the most sense for explaining how they spread across a large area in a short period of time, the older cultures in the region essentially disappear from archaeological record, and Steppe ancestry is at over 60% across Corded Ware range.

Samuel Andrews said...

It is interesting how no population with ~70% Steppe ancestry survived in Corded Ware zone even though even almost all Corded Ware samples have around that much.

Swiss Late Corded Ware is ~60% Steppe, Sweden Battle Axe is ~60% Steppe, very late Corded Ware Sweden is ~55% Steppe, Bell Beaker who is late Corded Ware is 45-60% Steppe.

Maybe, the admixture process between Steppe and farmers (and hunter gatherers) continued till 2300/2000 BC when Steppe ancestry dropped to modern levels of 50-60%.

Most of the admixture happened in first 200 years. But, Steppe ancestry seems to have continually dropped for another 300 years till it plateaued at 50-60%.

Huck Finn said...

I wonder to what extent this so called "Uralic cline" is real, cf. for instance:

"With some exceptions such as Estonians, Hungarians and Mordovians, both IBD sharing and Globetrotter results suggest that there are detectable inter-regional haplotype sharing ties between Uralic speakers from West Siberia and VUR, and between NE European Uralic speakers and VUR. In other words, there is a fragmented pattern of haplotype sharing between populations but no unifying signal of sharing that unite all the studied Uralic speakers."

https://genomebiology.biomedcentral.com/articles/10.1186/s13059-018-1522-1#Fig3

If anything, the unifying theme might be related to very old, pre Bronze Age West Siberian i.e. ANE based features unifying the so called Forest-Tundra belt. To what extent it can be called Uralic, is a different question. In other words: I wouldn't personally hold my breath while waiting for the ancient but modern Nganasan like population to show up. Or ancient but modern Ket like, who may represent the other side of the same coin.




"

JuanRivera said...

Found some archeological information on the steppe. In 'Europe's first farmers,' page 82, among other interesting things, it's shown that cultures in the Neolithic North Caspian, Don-Volga Steppe, and North Caucasus Steppe were in the availability phase, and that it lasted in the latter two until the arrival of Yamnaya. In part of the middle Volga region, it apparently jumps directly from hunter-fisher-gatherers to 'farmers' (most likely herders) of the Agidel culture without availability and substitution phases. A nameless culture or phase apparently jumps from hunter-fisher-gatherers to the substitution phase without the availability phase. Sredny Stog is classified as substitution phase, while Novodanilovka is classified as 'farming'. Middle Dnieper is classified as substitution phase.

Norfern-Ostrobothnian said...

@Davidski
Speaking of Bolshoy Oleny Ostrov, would you call this a good estimation?
I used Sidelkino HG for 50% of the ancestry.

RUS_Bolshoy_Oleni_Ostrov_no_Sidelkino_HG,0.0641964 ,-0.3211108,0.1060462,0.0183466,-0.1243304,-0.055667,0.011375,0.0187376,0.0040908,-0.0196806,0.0553424,-0.0094418,0.0142422,-0.0560674,-0.0328168,-0.004508,-0.0035208,0.0073218,0.0026142,-0.0145066,0.0079104,0.001039,0.0074196,-0.0039514,0.0028976

vAsiSTha said...

@slumbery
Those distances look very high.

Davidski, how many snps does the transversions only model end up using?

Hodo Scariti said...

@ Samuel Andrews

There is only a problem: how to link Corded Ware to Bell Beaker. And, obviously, how to link Corded Ware to R1b.

Steven said...

Is it possible that Eastern Hunter Gatherer ancestry can be confused as Yamnaya ancestry?

gamerz_J said...

@Samuel Andrews

"Corded Ware ancestry is not unique to them though, every European ethnic group has significant Corded Ware ancestry."

Even Italians, Greeks and Spanish?

gamerz_J said...

@Davidski

Swedish and Norwegians don't have that much of Uralic ancestry, do they? Say their actual Siberian do be around 3-5%?

Archi said...

@Samuel Andrews

"The old IE languages of Southeast Europe, most importantly Greek, don't come from Corded Ware."

Greek also comes from the CWC. The Central European CWCs of the Proto-Unetice region migrated through the Middle Dnieper CWC to the territory of the Catacomb culture in the Don region, where they formed the Babino (KMK) culture. Its western branch on the Danube is the Proto-Greeks.

"For most part Balto-Slavs are a product of Eastern Corded Ware"

Eastern CWC is Fatyanovo, as you know it is not an ancestor to the Balto-Slavs.

weure said...

'It is possible to add Baltic hunter-gatherers to a similar model after dropping CWC_Baltic_early in favor of closely related samples from the Early to Middle Bronze Age Pontic-Caspian steppe.'

Lay man attempt with G25 scaled.

https://www.mupload.nl/img/cr67mxpbg.37.18.png

I guess also a HG hight along the inland North Sea Coast (TRB West was most probably especially high in HG).

Davidski said...

@Steven

Is it possible that Eastern Hunter Gatherer ancestry can be confused as Yamnaya ancestry?

It's not possible, because Yamnaya is ~50% EHG.

That leaves a very large ratio of non-EHG ancestry that needs to be accounted for, and that can only be done with Yamnaya or closely related populations.

Davidski said...

@vAsiSTha

Davidski, how many snps does the transversions only model end up using?

Around 150,000 depending on the samples.

But the statistical fits and distances in qpAdm and nMonte won't always show a close correlation because of recent drift in the Global25.

So even though they often show very similar ancestry proportions, they're actually testing somewhat different things. That is, deep genetic structure versus more recent genetic structure.

The lack of specific drift in the reference samples is reflected as a high distance in nMonte.

And the more reference samples that you use in nMonte, the less likely you'll get accurate ancestry proportions, because nMonte will try to lower the distance in any way it can.

Samuel Andrews said...

@Hodo Sacriti,
"There is only a problem: how to link Corded Ware to Bell Beaker. And, obviously, how to link Corded Ware to R1b."

Bell Beaker is derived from Corded Ware.

Aesch25
https://eurogenes.blogspot.com/2020/04/aesch25.html

Linderholm 2020
Corded Ware cultural complexity uncovered using genomic and isotopic analysis from south-eastern Poland
https://www.nature.com/articles/s41598-020-63138-w

Mittnik 2019
Kinship-based social inequality in Bronze Age Europe
https://science.sciencemag.org/content/early/2019/10/09/science.aax6219

Anthony Hanken said...

Kerttu Majander recently gave an online talk about her upcoming paper,

Genome-wide ancient DNA investigation of Eneolithic individuals from southwestern Russia reveals a Genetic contact point between the forest-steppe and steppe populations (Majander et al. 2020).

Did anyone catch this presentation? Apparently she talked a little bit about ancient Uralic DNA but I don't know the details.

The abstract for this study suggests Uralic languages may be linked to Native American like ancestry on the forest-steppe. It remains to be seen how they came to this conclusion...

Davidski said...

@Anthony Hanken

The abstract for this study suggests Uralic languages may be linked to Native American like ancestry on the forest-steppe. It remains to be seen how they came to this conclusion...

They made some bad assumptions and ran a few bad models to back them up.

There's no N-L1026 in their forest/forest-steppe samples. They're all R1 and Q.

Archi said...

@Samuel Andrews

Bell Beaker is not derived from Corded Ware.

Mike said...

@Anthony Hanken

Could you post the link to the online talk?

Rob said...

@ Sam

In those studies you mentioned; the cases of R1b appear to come from more atypical -styles burials.

Samuel Andrews said...

@Archi,
"Eastern CWC is Fatyanovo, as you know it is not an ancestor to the Balto-Slavs."

That's true. It isn't known exactly which Corded Ware group created pre, proto-Balto Slavs.

"Bell Beaker is not derived from Corded Ware"

Bell Beaker is Western Corded Ware.

Davidski said...

@Rob

There may have been different populations and sub-cultures within the Corded Ware horizon defined by somewhat different cultural traits and Y-haplogroup frequencies.

But there's no way that western Corded Ware (Single Grave) was rich in R1a rather than R1b.

Anthony Hanken said...

@Davidski

That makes more sense from sites in southwestern Russia. I was still hoping for more Uralic aDNA.

@Mike

As far as I know the talk hasn't been posted anywhere. You had to contact first to be provided a link.
https://cifu13.univie.ac.at/programme/pre-congress-2020/pre-congress/

Rob said...

The other thing is it’s unlikely that KMK is from central CwC . Seems more like a fusion of Catacomb & Srubnaja groups

Vladimir said...

@Anthony
@Davidski
Here are two links to this work. Please note that there is no phrase about Ural populations on the Oxford site at all. In the second link, this phrase looks somewhat conjunctural, since the work is presented at a Symposium on Finno-Ugric languages

https://app.oxfordabstracts.com/events/653/program-app/submission/123334

https://cifu13.univie.ac.at/programme/pre-congress-2020/pre-congress/majander-al/




Davidski said...

@Vladimir

It looks like they updated their abstract.

So we can expect them to also claim in their paper that their samples are from early Uralic and Indo-European speaking populations.

Although there's a chance that they only modified their abstract specifically for the Uralic convention, and the paper won't have anything about Uralic languages in it, especially after it clears peer review.

But these scientists are from the Max Planck Institute, so anything is possible. They probably don't even know what N-L1026 is.

Norfern-Ostrobothnian said...

@Davidski
Was the mention of Siberian ancestry added? Anyways if it's southwest Russia I don't expect anything north or east of Moscow, maybe Samara could be the eastern most boundary. But definitely should not bring up any Proto-Uralic people unless it is Volga-Kama spillage to the region. But we have to wait and see. If it were up to me the Turbino site should be the top target for Uralic ancient DNA in Europe.

Davidski said...

@Norfern-Ostrobothnian

This paper has samples from around the Urals.

But none of them have anything to do with the Uralic expansion, because they're too early and don't have the type of Nganasan-like Siberian ancestry that matters in this context.

The contact zone between early Indo-Iranians and Uralians was obviously the place where R1a-Z93 and N-L1026 met, and yeah, that was the Seima-Turbino network.

Huck Finn said...

@ Anthony, @ D and re: "The abstract for this study suggests Uralic languages may be linked to Native American like ancestry on the forest-steppe. It remains to be seen how they came to this conclusion..."

"They made some bad assumptions and ran a few bad models to back them up."

I think the conclusion is most logical, based on previous studies. The only thing autosomally uniting Uralic speaking groups, if even that, is the ancient ANE based genetic heritage of West Siberia, nowadays poorly visible in modern Forest Tundra belt. It may then be that paternal N was first not anyhow related to that, we'll see.

Norfern-Ostrobothnian said...

@Davidski
Upper Kama region had contacts to West Siberia. So how old are we talking and what regions exactly? Do remember that Nganassans aren't pure Uralics either. And also if it's Seima-Turbino network as the contact, and the Turbino site is continuous with Garino-Bor culture...

Davidski said...

@Huck Finn

It may then be that paternal N was first not anyhow related to that, we'll see.

I can already tell you that it wasn't.

And yet, somehow Baltic Finns, Hungarian conquerors, Nganasans and many others ended up with a lot of N-L1026, and at least some Nganasan-related genome-wide admix.

Norfern-Ostrobothnian said...

The subclades of N are so overlapping among Uralics that I would still bet some common root, at least West of the Urals. If not, then there would be roughly 3 groups with Finno-Mordvinic, Mari-Permic and Ugro-Samoyedic, with the Saami being an outlier genetically (although their relatives once inhabited territory conjoined with Permic and Mari ancestors in Arkhangelsk region).

Davidski said...

@Norfern-Ostrobothnian

West Siberia around 2,000 BCE is where you'll first see N-L1026 and Nganasan-related ancestry.

Norfern-Ostrobothnian said...

@Davidski
Barabat valley right?

Norfern-Ostrobothnian said...

What culture(s) on this article would you connect with N-L1026 and/or Uralic peoples?
https://nashural.ru/article/istoriya-urala/doistoricheskie-vremena-zauralya/?amp

Davidski said...

@Norfern-Ostrobothnian

Baraba steppe.

I honestly don't know what might be securely classified as Uralic before 2,000 BCE. Seima-Turbino was probably only partly Uralic.

Copper Axe said...

What is up with this general pushback about Uralic languages hailing from East Siberia? Seems rather obvious based on the overlapping ancestries of Uralic peoples in my opinion.

Copper Axe said...

It is best not to think of Seima-Turbino as a culture with a distinct identity but as a trade network involving many different ethnicities. It is called a phenomenon after all.

Uralic speakers fit into this somehow but whar their exact role was unknown. Indo-Iranians were in there as well plus all the other ethnicities along the trade network. Hell, maybe Tocharians were involved in the Seima-Turbino as well, it reached further into China than Tocharians did plus there is this linguistic article which suggested contacts between Tocharians and Samoyeds.

Archi said...

Krause, as always in his repertoire, makes unfounded statements. He wrote "the earliest detected occurrence of mixed EHG and Iranian Neolithic-related ancestry (previously described as ‘steppe ancestry’).", but for sure there will be no mention of CHG.

Let's see if there is at least some justification for his claim.

Norfern-Ostrobothnian said...

Krotov culture of central Siberia northwest of the Okunev culture was involved, being pretty close to the Baraba steppe. On the other hand Turbino site lies outside of Andronovo or Balanovo territory.

Matt said...

Quick comparison of the two-way models proportions to a G25(scaled)+Vahaduo model with exactly the same source individuals (I can't be sure if exactly the same individuals are present in targets, though): https://imgur.com/a/5hQuvpV

There is a perfect correlation for populations that are mostly on the Baltic-Slavic cline (Czech, Slovak, Polish, Ukrainian - remarkable how 1:1 the proportion is), while the populations in NW Europe deviate towards having slightly more Middle Neolithic European ancestry on Global 25 data. All in all though its pretty much in the same estimated range of 50-60% range of ancestry from the three CWC individuals Davidski selected, between two models with the exact same sources.

Those CWC samples in turn have a mean proportion of 92% Steppe_EMBA or 82% Steppe_Eneolithic (Piedmont) ancestry under similar G25 models (X+SWE_TRB), so that would I guess equates roughly to about 45-55% Steppe_EMBA or 40-50% Steppe_Eneolithic ancestry, in the populations set in the two-way fits, today. "Steppe ancestry" is a bit of a moving target at these relatively fine levels.

....

@Davidski, re; I haven't really read anything about any examination of whether f3 stats are affected at all by recent drift or not, really; it's not like its implausible (e.g. we know sometimes closely related populations do pair in f3 outgroups, like the British Neolithic with each other, or the Globular Amphora from Poland and Ukraine with each other), though the influence of such should be relatively reduced compared to other statistics.

Whether f4 stats and D-stats have reduced influence from any recent drift relative to f3 and this affects patterns of relatedness by Northern Europeans to different Southern European populations in f3 stats, could be tested by taking something like (A,B;C,Outgroup), where A and B are Western and Eastern European populations with the same formal proportions of underlying ancestry in your models and then C is a Southern Europe. E.g. (Czech, French_Northwest;Spanish,Mbuti). To get really precise you could ignore the population labels of different individuals on Eastern and Western European clines and just assemble sets that have alike degrees of ancestry proportions. Based on some of the methods used by Lazaridis, I suspect that would more precisely "lens" exactly the differences in relatedness, once accounting for proportions.

@Samuel, It is interesting how no population with ~70% Steppe ancestry survived in Corded Ware zone even though even almost all Corded Ware samples have around that much.

It's perhaps not so surprising though; where we have studies in the Baltic and Scandinavia (Saag, Margaryan) that proceed through the Iron Age->medieval, it looks like there's low-level ongoing northward migration from at least as far as Central Europe which leads to some population change (in the Baltic this seems to be in concert with either dissolving, or reducing to a very lower and statistically harder to detect level, those differences in ancestry between autosome and X-chromosome, so may possibly have been male biased enough to diminish earlier signals).

Davidski said...

@Matt

It says here that f3 stats can be impacted significantly by recent drift.

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3522152/

And I think it's pretty obvious that the plots you posted do reflect recent genetic drift.

You would have to get in touch with Nick Patterson to ask about whether f3 stats are more sensitive to recent drift than f4 and D stats, but I'm pretty sure that they are.

In any case, it's not something that I want to test.

weure said...

@Archi
'Bell Beaker is not derived from Corded Ware.'

At least for NW Europe, specific the Dutch/Rhenish Beakers could this well (SGC>BB) be the case.

In this area we see a cultural evolvement to the BB package, without new genetic influxes c.q. major immigration.

Archi said...

@weure
"At least for NW Europe, specific the Dutch/Rhenish Beakers could this well (SGC>BB) be the case.
In this area we see a cultural evolvement to the BB package, without new genetic influxes c.q. major immigration."

We see a mixture of BB and CWC there. This is where migration is clearly recorded.

@Samuel Andrews

Bell Beaker is not Western Corded Ware.

Davidski said...

@Archi

Dutch Bell Beakers are identical to German Corded Ware except with some local hunter-gatherer ancestry.

So the only migration into the region during the Late Neolithic was the Corded Ware migration.

Archi said...

@Davidski

"Dutch Bell Beakers are identical to German Corded Ware except with some local hunter-gatherer ancestry."

How do you understand? What were they R1a?

"So the only migration into the region during the Late Neolithic was the Corded Ware migration."

In all sciences, this is impossible.

Davidski said...

@Archi

You'll have to eventually accept that Bell Beakers are derived from western Corded Ware.

There are already enough samples to be confident about this conclusion, and more are on the way soon from the Netherlands. :)

Matt said...

Yes, that paper states admixture f3 statistics can be masked out by *very* high levels of population specific drift, probably Kalash or Native American sized effects. As to whether the patterns in f3 statistics reflect a uniform population experiencing subsequent drift, or differing substrates, I don't know and don't think anyone does. You're free to test what you want of course, I'm only proposing a way you could back up the statement, you can do so or not depending on your own interests.

Rob said...

I’m can’t see why people still going about Dutch beakers
The relevant studies showed that BB entered corded Ware during the middle phase
Maybe PFB evolved in the rhine land; but that’s a separate issue

Davidski said...

@Rob

The relevant studies showed that BB entered corded Ware during the middle phase.

From where?

The southeast Polish Corded Ware R1b samples aren't closely related to Bell Beakers, nor are they ancestral to them.

R1b was already near the Rhine much earlier than these samples.

Davidski said...

@Matt

Well, if f3 mixture stats can be significantly skewed by recent drift, then f3 stats aren't immune to recent drift, and it's likely that a plot made from f3 stats will reflect recent drift.

The claim you made in your initial comment was that since your plot was based on formal stats then recent drift wasn't an issue. But it is an issue always, it's just that it's a bigger issue in, say, PCA.

weure said...

Simple even nowadays the Veluwe area, heartland of the Veluvian Bell Beakers in the middle of the Netherland just above the Rhine, shows a relative higher R1b P312. R1b P312 is also the Y-DNA of the BB samples of West-Friesland (Olalde).

In archeology you can see that 2800 BC there was a shift in the North Dutch area towards Single Grave c.q. the Steppe Pastoralist came in. Afterwards there was only a cultural shift toward the Bell Beakers no signs of immigrations.

Archi what is your counter narrative with respect to the Dutch Beakers?

Anthony Hanken said...

@Davidski

"This paper has samples from around the Urals."

Wow so N was even absent around the forest/forest-steppe Urals during the Eneolithic. Thats certainly a testament to a rapid and late expansion.

If S-T spread N-L1026 and Uralic languages and their modern day representatives do not have significant R1a-Z93 then what role did the Indo-Iranians play in the S-T expansion?

@Copper Axe

Seima-Turbino was a trade network but local metallurgical centers/siberian style pottery as well the proposed genetic and language shift suggest a demic trans-cultural population movement as well as trade.

If you have a translaor plugin I would suggest checking out this website. Please note some stuff is outdated but pay attention to the dates cited, there is still tons of useful information.
https://arheologija.ru/pamyatniki-seyminsko-turbinskogo-tipa-v-evrazii/

Nick Patterson (Broad) said...

My general advice for qpAdm is
1) Work on the right hand set.
Don't include irrelevant population
(except for one population as an outgroup);
picking the best RHS can dramatically
reduce s. errors on the admixture weights.
2) If qpAdm gives a very low p-value
try and understand why, sometimes
it is telling you that the target
is not a mixture of the sources but
sometimes the assumptions are violated,
for example recent gene-flow from
left pops -> right.

Samuel Andrews said...

Corded Ware was not a single R1a M417 population. It was a composite of multiple Yamnaya tribes who carried various clades of R1b L51 and R1a M417.

Samuel Andrews said...

We used to think Bell Beaker can't possibly be from Corded Ware. Then we learned it is from Corded Ware.

We now think Corded Ware can't possible be from Yamnaya. I think we will learn Corded Ware is from Yamnaya.

We ade these wrong predictions, because understood IE genos were composed of a single Y DNA and IE cultures correlate with IE genos. But we were wrong about IE cultures being entirely composed of a single genos.

Sredny Stog, Yamnaya, Corded Ware were all composed of multiple genos belonging to multiple R1b M269 and R1a M417 lineages.

Yanaya comes from Sredny Stog, Corded Ware comes from Yamnaya, Bell beaker comes from Corded Ware.

Yamnaya/Sredny Stog/Corded Ware spoke Late Proto Indo European.

Rob said...

@ Davidski
The earliest elements of the BB package form in southern Poland . The path of L51 tracks via southern Poland, southern Germany and Switzerland. All early CWC in Germany; Baltic, Sweden lack R1b; which warns against the Baltic -Dutch route

Rob said...

From the suppl in linderholm et al

“ In comparison with other regions of Central Europe, the Małopolska burials are equipped with particularly rich sets of gifts.
Particularly specific is the nature of the equipment in some graves of adult men, which include
ceramic vessels (the main type is a large beaker with a distinct neck), weaponry (stone battle-
axes and arrowheads), sets of flint tools (mainly axes, knife inserts and strikers), bone, antler
implements, stone implements (whetstones and grinding discs), copper tools and ornaments (including characteristic copper hair-rings), as well as “half-finished” items implements of flint. Thus, such burials emphasize the role of a man as a warrior, as well as a craftsman-specialist, whereas in the Małopolska materials the specialization in the field of flint making is especially emphasized. Unlike in other Central European CWC groups, at Małopolska there are numerous burials of men equipped with triangular arrowheads (up to 30 pieces). In other areas, the archerysets as an important element of the burial equipment of the deceased appears about 100-200 years later associated with the appearance and spread of the BBC ritual.”

Davidski said...

@Rob

Here's another quote from that paper:

The PCA revealed that despite geographical proximity there is a distinct genetic separation between CWC and BBC individuals from southern Poland. The genetic variation of CWC individuals from southern Poland overlaps with the majority of previously published CWC individuals from Germany while the eight published CWC individuals from the Polish lowland [10,11] more closely resemble BBC individuals (Fig. S21). This fact is not unexpected if we consider the CWC communities in Polish lowlands as representatives of north-western parts of the CWC world called as the Single-Grave culture (see supplementary information). The genetic variation of BBC individuals from south-eastern Poland overlaps with the broad variation of BBC individuals from Central Europe (Bohemia, Moravia, Germany, south-western Poland and Hungary) (Fig. S22) which corresponds well with archaeological data.

Bell Beakers moved into Poland from the west, and Single Grave is implicated here.

Of course there are Single Grave samples on the way, including from this team.

Archi said...

@Samuel Andrews
"We used to think Bell Beaker can't possibly be from Corded Ware. Then we learned it is from Corded Ware."

We did not learn this. No slightest evidence of this has appeared. This is only a hypothesis spread by the author of this forum.


"We now think Corded Ware can't possible be from Yamnaya. I think we will learn Corded Ware is from Yamnaya."

You are wrong again.

Davidski said...

@Wise dragon

Opinions about physical anthropology and looks are against the rules at this blog.

They're too subjective and too far removed from modern science.

Please stick to genetics.

Rob said...

@ Davidski
That doesn’t disprove my contention because the R1b guys in their study were classed as “CWC”; despite being culturally distinctive
There was then a later Central European BB movement into Silesia
So; protoBB moved via southern Poland followed by classic BB back eastward

Davidski said...

@Rob

Yes, I know, but I don't agree.

I think that the Bell Beaker population was derived from the Singe Grave population.

Archi said...

@Davidski

It just goes to show that the early Bell Beakers were incorporating into the CWC communities, bringing their own traditions while still taking over the CWC traditions. Syncretic groups were formed. In the Malopolska groupe this is exactly the kind of group that deviates from the CWC traditions extremely strongly because they are Bell Beakers by origin.


@Samuel Andrews

"We used to think Bell Beaker can't possibly be from Corded Ware. Then we learned it is from Corded Ware."

We did not learn this. No slightest evidence of this has appeared. This is only a local hypothesis.

"We now think Corded Ware can't possible be from Yamnaya. I think we will learn Corded Ware is from Yamnaya."

You are wrong again, there is not the slightest proof of this.

Wise dragon said...

@ Davidski,

Okay. I didn‘t know that it's against the rule since in genetic papers they discuss phenotype too. Sorry.

Samuel Andrews said...

@Archi,

Aesch25 R1b L151 predates Bell Beaker by 200 years. It is anarchistic to say he descends from Bell Beaker.

Where did Aesch25 come from? Well, the only Steppe people living in/near Switzerland in 2700 BC was Corded Ware. The best option therefore is Corded Ware. And it makes sense considering Aesch25 is genetically identical to Corded Ware main cluster.

Plx, do try to explain how Aesch25 isn't Corded Ware.

Samuel Andrews said...

Archi, I remember when we did not know Aesch25's radiocarbon date you argued he will date to the Bell Beaker period. You claimed he came from a Beaker burial.

You turned out wrong.

Davidski said...

@Wise dragon

Some phenotypic traits are indeed discussed in genetic papers, but very specifically, in the context of the markers that are known to be associated with them.

No genetic papers have yet discussed the fact that Bell Beakers were often brachycephalic while Yamnaya and Corded Ware people mostly dolichocephalic, even though their ancient ancestry was similar.

It might have been due to recent selection, genetic drift, diet and/or epigenetic factors. But whatever was the cause, there are no specific markers or factors that can be linked to it yet.

So it's not like discussing eye color, hair color or even height. It's much more subjective and can't be backed up with any hard science yet.

Davidski said...

@Archi & Samuel

Well, there's Aesch25, ALT-4, the French Corded Ware R1b, and the forthcoming Single Grave L51 from a burial that has nothing to do with Bell Beakers.

And within the next 12 months we'll have a lot more evidence that L51 was sitting near the Rhine since ~2,800 BCE.

So save your energy. This is already a pointless discussion.

FrankN said...

Rob - on BB:

The genesis is now becoming pretty clear - introduction of the Single Grave Burial Rite (look it up with Furholt) into Western Neolithic Cultures such as the Artenac and epi-/post-SOM. Moreover, thanks to recent French research, we now also have a pretty good timetable on SGBR BB burials in NE France. The earliest SGBR BB burial known so far all across Europe is Ciry-Salsogne - and we even have aDNA from there!

Further reading (in French):
https://www.persee.fr/doc/pica_1272-6117_2011_hos_28_1_3323
https://www.persee.fr/doc/galip_0016-4127_2005_num_47_1_2047#galip_0016-4127_2005_num_47_1_T1_0111_0000
https://www.researchgate.net/publication/302942778_Chronologie_et_facteurs_d%27evolution_des_sepultures_individuelles_campaniformes_dans_le_Nord_de_la_France

I have put some of the places mentionned there, together with those equally early Belgian/Luxembourg sites, and some places along the Rhine where we have aDNA from, on a map (select pedestrian mode for a better idea then it should be self-explaining):
https://www.google.de/maps/dir/Oostwoud,+Niederlande/Mol,+2400,+Belgien/W%C3%A9ris,+6940,+Belgien/Altwies,+Luxemburg/Achenheim,+Frankreich/H%C3%A9genheim,+68220,+Frankreich/Longvic,+21600,+Frankreich/Jablines,+Frankreich/Ciry-Salsogne,+02220,+Frankreich/Mol,+2400,+Belgien

Otherwise, of course, BB was a polycultural and in all likelyhood also polylinguistic phenomenon. Within it, the SGBR BBs can in all likelyhood be regarded as IE-speaking - that also includes most of British BBs, which archeologically have lots in common with BBs from N. France (albeit I wonder whether we have to consider a later, ca. 2,000 BC, overforming from Lower Saxony and the Middle Elbe). Elsewhere, especially Iberia, the Midi and most of Burgundy, where BBs are for a long time dominated by collective burials, I have my doubts about the prevailing language. SGBR in earnest appears in Iberia only with El Argar, and that shift in burial rite corresponds with R1b becoming the predominant yDNA.

Rob said...
This comment has been removed by the author.
Archi said...

@Samuel Andrews

"Aesch25 R1b L151 predates Bell Beaker by 200 years."

Don't tell a lie, the dating of Aesch25 2864-2501cal BC fits well with the BBC dating.

Don't be fake, Aesch25 has been attributed as Bell Beaker, proven. This is an archaeological fact. He is exactly the BBC, this is a scientific fact and his dating has nothing to do with it. Nobody can attribute it as CWC, it is impossible.



weure said...

@Davidski:
'No genetic papers have yet discussed the fact that Bell Beakers were often brachycephalic while Yamnaya and Corded Ware people mostly dolichocephalic, even though their ancient ancestry was similar.'

Het I can recommend the work of Kurt Gerhardt (1976), he has spent his life time to study the physical anthropology of the Bell Beakers.

He made clear that there is a development of the robust-dolichomorph that he finds characteristic of the West-German Corded Ware (=Single Grave like) to the most typical skull type of the Bell Beakers (their 'marker')!

This evolved from the robust dolicho (SGC) by two developments:
- brachcephaly
- hypsicephaly

....into the BB Steephead.

You see this development in type 1, by Gerhardt 1976. Long time ago but I guess his conclusions are not altered.

https://www.mupload.nl/img/1ug9c1.jpg

Apostolos said...

I think that both the ancestors of the R-M417 and the R-M269 guys were hunter-fishers, settling all over Europe, parts of Northern and Central Asia and even West Asia, near large bodies of water.

It is important to wonder where they adopted stockbreeding, how and if that was the result of two separate events.

By the way, one advantge CWC people seem to have had was their diversified diet, which included fish.

They had sheep, cattle, horses, dogs (which could have been used for hunting at least occasionally), at least some knowledge of agriculture and they were eating fish too.

I personally think there were two separate events.

epoch said...

The Seine-Oise-Marne culture has quite an interesting connection to Dutch SGC. Grande-Pressigny flint knives are recovered from Dutch SGC and BBC barrows and there is a cave burial 50 km from Grande Pressigny where a battle axe was recovered. From what I get GP is connected to SOM.

Brittany is a hotspot for Maritime Beakers associated with communal Megalith burials. If these are really farmer cultures as the original Iberians - I know there is discussion if this is accurate but let's bear with this - then maybe the maritime beakers found with fully farmer beaker cultures are imitations because of cultural transfer.

If these beakers are connected to alcohol (beer) than that could be what the farmers took up.

Davidski said...

@Apostolos

I think that both the ancestors of the R-M417 and the R-M269 guys were hunter-fishers, settling all over Europe, parts of Northern and Central Asia and even West Asia, near large bodies of water.

Which data do you base this statement on?

Can you show me the hunter-fisher samples from West Asia belonging to R-M417 and R-M269, or even R1a and R1b?

Davidski said...

@All

I've updated the blog post with Nick Patterson's advice about qpAdm.

Rob said...

@ Epoch

''Brittany is a hotspot for Maritime Beakers associated with communal Megalith burials. ''

Its the very same misconception
Beaker goods were not placed in the communal part of the chambers, but at the entrance which had been blocked off by a boulder. Usually there is a later, Beaker-man burial in that spot.

E.g. from the same author which Frank mentioned
Look at Fig 5 (a true communal burial from pre-Beaker era) versues Fig's 6 - the re-use phase
https://hal.archives-ouvertes.fr/hal-01314028


As I said, it seems to be the same scenario everywhere in the so-called Southwest domaine
The only question is why Maritime Beakers were so favoured in Brittany over other styles

Archi said...

Davidski said...
"Nick Patterson's advice about qpAdm. My general advice for qpAdm is 1) Work on the right hand set. Don't include irrelevant population (except for one population as an outgroup); picking the best RHS can dramatically reduce s. errors on the admixture weights. 2) If qpAdm gives a very low p-value try and understand why, sometimes it is telling you that the target is not a mixture of the sources but sometimes the assumptions are violated, for example recent gene-flow from left pops -> right."

Some kind of nonsense. If you can even have a hypothetical gene flow from left populations to right populations, then your model is completely wrong. By definition, right-wing populations are outgroups, that is, they must be as unrelated as possible to the tested populations, older than them and as far from them as possible.

All tested variants should be added to the left populations, since it is from them that the best variant should be chosen. Comparable p-values ​​exist only in the current test, from different tests the p-values ​​are not comparable, since they depend on the number of left populations and at the current right populations. With a different number of left populations, the p-values ​​cannot be compared at all, they are simply calculated differently.

leftpops1 A,B,C
p-value1 001
leftpops2 A,B
p-value2 00
p-value1 != p-value2

Davidski said...

@Archi

OK, if you say so.

I look forward to seeing your own program that does a better job than qpAdm.

But I'll probably be waiting forever.

Archi said...

@Davidski

It's not about whether qpAdm is good or bad, but about how to use it, what to input to get a correct output for comparison.
Don't worry about me, I have my own program.

Garvan said...

@Davidski
While thinking about what should be in right populations, I wondered if the left population Baltic_LTU_Narva (7800 BP) is too close in time and place to the right population RUS_Karelia_HG (7200 BP) for the model where you tried HG in Polish. Do I have the dates correct?

Davidski said...

@Garvan

Karelia_HG is way older than Narva, which is "Neolithic". You can check their dates here...

https://reich.hms.harvard.edu/downloadable-genotypes-worlds-published-ancient-dna-data

Apart from that, Narva is a mixture of WHG and Karelia HG (EHG).

vAsiSTha said...

@archi

"By definition, right-wing populations are outgroups, that is, they must be as unrelated as possible to the tested populations, older than them and as far from them as possible."

They can be related. A right pop can be a source to the left pop.

Eg. Yamnaya in left pop and karelia_hg in right pop is acceptable, but not the other way around.

This is because there is no way there is any geneflow from yamnaya to karelia_hg, karelia having 0 chg/Iran.

I have seen your qpAdm models, they suck.

vAsiSTha said...

@archi

"Comparable p-values ​​exist only in the current test, from different tests the p-values ​​are not comparable, since they depend on the number of left populations and at the current right populations. With a different number of left populations, the p-values ​​cannot be compared at all, they are simply calculated differently."

Incorrect again.

Look up the rotating model in Harney et al preprint (which is recommended as gold standard to identify most optimal model), in which p values are compared across models with diff sources as well as right pops.

WeightofAudio said...

@Davidski,

If you could please; what is the consensus now regarding Narva admixture in NE Europeans? Modest admixture?

Thank you

Archi said...

Davidski, why didn't you miss my response to vAsiSTha's mistakes? I answered him immediately.

@vAsiSTha

"They can be related. A right pop can be a source to the left pop."

Outgroup does not mean that in the distant past he could not be an ancestor, this is a very controversial issue, because in testing we don’t know who in what way influenced the tested population. The main thing is that the right population belongs to the same group as the left populations. Not being in a group is not being in a relationship. The classic outgroup scheme is as follows https://i.ibb.co/q9713mS/F4statistics.png

"Look up the rotating model in Harney et al preprint (which is recommended as gold standard to identify most optimal model), in which p values are compared across models with diff sources as well as right pops."

I wrote it right, you wrote it wrong. Cannot be compared between different rotating right populations, right1 != right2, can only be compared when right1 = right1 and right2 = right2. There cannot be a different number of sources in left populations, I emphasize once again - numbers, and not just different left populations. You did not understand anything in the papers, your statement is not written there, you are inventing, the authors do not contradict the rules of use qpAdm.

All your qpAdm models are completely wrong, you do not know how to use this program.

quote from Harney et al. 2020
"1) Ensuring that the model includes right populations that are differentially related to the
various source populations
that are being used as potential left populations.
2) Ensuring that models are directly comparable. It is not appropriate to compare two
models that use entirely different sets of right populations.
While it may not be possible
to use identical sets of right populations for all models under consideration, the right
population sets should be as similar as possible."

You didn’t even pay attention to the values ​​that are obtained at the output of this program, you just don’t understand them, just as you don’t understand what method they are considered. It is elementary to prove that you are wrong by simply running the program.

left pops:
Russia_EBA_Yamnaya_Samara
Russia_EHG
Georgia_Kotias.SG
00 0 9 3.293 0.951558 0.378 0.622
01 1 10 10.536 0.39475 1.000 -0.000
10 1 10 5.804 0.831449 0.000 1.000
https://pastebin.com/4s6DAYY3

left pops:
Russia_EBA_Yamnaya_Samara
Russia_EHG
Georgia_Kotias.SG
Ukraine_Globular_Amphora
Russia_Piedmont_Eneolithic
Ukraine_N
00111 3 9 3.162 0.957518 0.378 0.622 -0.000 0.000 -0.000
01111 4 10 10.112 0.430681 1.000 -0.000 0.000 -0.000 -0.000
10111 4 10 5.695 0.840169 0.000 1.000 0.000 -0.000 -0.000
https://pastebin.com/7eTBJeUA

vAsiSTha said...

Lol Archi, you are using the HO dataset without allsnps. So your models end up using just 139k and 96k snps respectively hahahaha.

Your std errors in both models are between 20% and 40% lol.

Noone should take your models seriously. or your understanding of how all these things work. These are the worst models i have ever seen.

1.) i dont think you are good at understanding english. You said "they must be as unrelated as possible to the tested populations, older than them and as far from them as possible."

No they in fact do not need to be 'unrelated' or 'as far away as possible'. What they need is

a) populations in right pops should not have shared drift with target which at least 1 of the sources does not share with target(lines 195-202 Harney et al 2020).
ie. None of the right populations are more closely related to the
target population than to either of the source populations.

b) You need some 'differentially related' pops in right pops list to have a good model to be able to differentiate between potential sources. 'Differentially related' does not mean 'unrelated', neither does it mean 'a far away pop'. It means quite the opposite actually. eg a pop from congo is useless in modeling swedish people (except as just 1 mandatory outgroup). This is exactly what Patterson says above in his comment.


You also said "from different tests the p-values ​​are not comparable, since they depend on the number of left populations and at the current right populations"

Harney's rotating model has different sources and different right pops and the p-values are compared across all the rotating models. You were wrong, period.

"An alternative to the “base” reference set strategy that has been implemented in order to avoid these problems is to create a set of “rotating” models in which a single set of populations is 347 selected for analysis (e.g. SKOGLUND et al. 2017; HARNEY et al. 2019). From this single set of populations, a defined number of source populations are selected, and all other populations then serve in the reference population set for the model. Under this “rotating” scheme, populations are systematically moved from the set of reference populations to the set of sources. Thus, all population models are generated using a common set of principles and are therefore more easily directly compared." (harney et al, lines 345-353)

As usual, this will be my last reply to you. Start reading how Patterson's papers model samples. Narsimhan et al is a good start. Look at his right pops for steppe samples.

Archi said...

@vAsiSTha
Lol @Archi, you are using the HO dataset without allsnps. So your models end up using just 139k and 96k snps respectively" hahahaha.
"Your std errors in both models are between 20% and 40%" lol.

You do not understand anything in the parameters, you have invented something for yourself and write complete nonsense. You write words that you don't understand. You do not understand questions at all, your models are always wrong because you do not know anything and everything that you do is just a pure fake for the result. Nobody takes your models seriously because they are pure deception and everyone knows that. Everything you wrote has nothing to do with the case.

"No they in fact do not need to be 'unrelated' or 'as far away as possible'. What they need is"

The absence of a common drift means being outside the group, that is, either being older or being far away. Since you do not understand this absolutely all your models are wrong.

https://i.ibb.co/1sMPKrw/qp-Adm-groups-model.png

"Harney's rotating model has different sources and different right pops and the p-values are compared across all the rotating models. You were wrong, period."

No, you do not understand the texts you quote. Comparison of values ​​is performed only within one Rotating Model, with matching sets of right values. You, as always, were wrong without exception, period.

The rotating method is their suggestion is not standard. But most importantly, it does not contradict what I wrote, but the quote you quoted contradicts your own statements. You, as always, did not understand the text.

Tigran said...

Will samples from NW Russia closer to the Urals also be R1a/R1b and EHG like?

Archi said...

That model used the Rotational Method for the experiment, so Iran_Mesolithic_BeltCave_all_published was added to the right populations, which shouldn't be there for this set of left populations. Here is the same data without this population.

00 0 8 10.674 0.220846 0.438 0.562
01 1 9 101.008 9.83632e-18 1.000 -0.000
10 1 9 63.995 2.26172e-10 0.000 1.000
https://pastebin.com/ftr4HjUS

00111 3 8 10.320 0.243277 0.443 0.557 0.000 -0.000 0.000
01111 4 9 99.246 2.23565e-17 1.000 -0.000 -0.000 -0.000 0.000
10111 4 9 65.119 1.36808e-10 0.000 1.000 -0.000 -0.000 0.000
best pat: 00111 0.243277 10.320 not nested
https://pastebin.com/uzHG3sdz

As you can see, there is a striking difference only when changing one right population! Therefore, it is impossible to compare the p-values ​​obtained with different sets of right populations. Of course, such a strong effect is due to that one of the right populations actually belonged to a group with one of the left populations. In no case should the same population be on the left and on the right, as some do. That is why there is no single set of right populations for all cases; some right populations can always enter a group with a left population. A strong dependence on right populations is the main drawback of this method. It is necessary to be very careful when choosing the right populations, it is best to use those published in the scientific literature, and not those that come up with the indecipherable vAsiSTha. The rotational method was generally invented only for separating close p-values; it is not the p-values ​​themselves that are important in it, but the dynamics of the change in p-values.

galadhorn said...

I hope it can be of interest for you:

aldrajch.blogspot.com/2020/09/10-archeogenetycznych-atutow-polski.html

I have used some information from you, Davidski.

I have made a kind of summary about the bright side of being a descendant of the PIE > Pre-Balto-Slavic > Pre-Slavic in Poland