G25 g25 VS qpadm/admixtools2 comparison.

[eupator]

The discourse has been stellarly upgraded. I read your efforts with great interest.

 

[Archetype0ne]

The real question is how would you model north Italians as Nausius and Slavic, your parameters are stretched to produce comical results.

That was the rhetorical question I think I went out of my way to help you understand. Here to make it plain for you, you don't!

Those parameters also model you as Illyrian plus Slavic, while my models are always within cline. This does show that qpdam is nowhere near full proof. Your Illyrian model is forced.

I am not sure what your models having a cline (all models do lol) have anything to do with this. And yes you can model Albanians with Illyrian + Slavic as did Lazaridis and even Olade and the Serbian academy did, whether you like it or not. Also enlighten me how is my model forced? I literally even removed all the unnecessary fluff from the Olade tail, the tail literally works with just 17 populations. This is the tail I had on hand and had to force nothing. Meanwhile to make your test work you had to adjust your tail for 2 days? With 24 populations, now that is forcing and overfitting a model my friend. You are going based on the result you "want" to find, and over fitting the model till it passes.

I did not claim Estonia_BA is Slavic., show me where I stated that. It has the Slavic drift, this was well known and discussed in Anthrogenica, where your favorite, Brumziu tried hard to argue against it, against some top-notch northern Slav who knew his stuff.

You did not?

Or did you?

And don't be condescending, take a look at your model, you stretched out the weights to produce unrealistic results.
My Slavic mdv values are better than yours. You know there are some unmixed Slavic samples among the Vikings samples right? Did you know that mr. research?

How did I stretch my model when you had to add 24 populations to your tail mate? Your tail has plain, .DG, .SG, single sample populations, aggregates and anything you can find. I went with Olade labels, and just cleaned up their tail per https://academic.oup.com/genetics/article/217/4/iyaa045/6070149
Guess "mr.research" makes me condescending? To spell out sarcasm you will not get, I was being constructive, your ad hominems give away your pettiness.
But either way, how are Baltic BA better proxies for Slavs when not only they lived thousands of years ahead of the Slavs, but outside of being connected to each other due to Baltic BA which contributed to the Baltic drift in Slavs, with the additional steppe they had creating the Balto-Slavic artifact, then samples like Czech_EarlySlav (nomenclature by Reich himself) or Av2, which we know for a fact lived at the very early stages of Slavic (proper) ethogenesis and shared early Slavic cultural complexes?
Slaivc samples among the Vikings? Do you still not get what asynchronous means? What is the date of AV2? 550-600. Czech_EarlySlav? 660-770. What about the Vikings? Use your logic the Slavic Vikings are the descendants of admixed populations of Early Slavs to Nords. And this population did not travel back in time to create the Slavic admixture in Balkan populations. I do not know how to explain it more plainly to you. If you don't get it, tough luck.

Just like G25, qpdam can be used as metal detector, and nothing else. The only way to truly prove it, is to have relevant samples every 500 years, where you go by the best fitting model. Such ideal dataset will not be available for the meantime.

The issue with qpdam is that it has hick ups when there is shared drift. In reality, the Alb parent population was largely IA but still had acquired some MENA admixture. And the additional MENA that came from Byzantine period had acquired some steppe admixture. These two proxies are problematic, qpdam wants to work with cleaner coordinates, that's why EST_BA passes, while Slavic_mdv fails, not because EST_BA is the real Slavic, but qpdam likes proxies with distances/distinction. My models are not non-sense. I kept my parameters free of bias.
For Qpdam to work the way you want it. You will need to find the parent Alb population mixed with MENA before Salvic admixture occured. With the current samples, I say good luck with that. And it's frankly never going to happen, because some MENA admixture occured before Slavs, and later it came with Balkan Slavs. You will not get the recontruction/model you desire, because of how the vriables were formed.

Hmm, I wonder where we could find a population like that... Lacking any sort of Slavic, and rich in Albania_BA_IA + Eastern / Anatolian
Did you even read the models I shared with you? With your own "proxies" lol.

How is your model stretched to get the reconstruction I desire when you "literally" came up with this garbage?
If I was to do it, it would look something like this.

You should try modeling kenete and shtike seperately, as they are not the same.



Baltic_BA was used as a Slavic drift proxy, you do not get that? You think my master plan was to show low Slavic admixture?
Salvic drift using EST_BA average in G25

Try mine.

right = c('Cameroon_SMA', 'Czech_Vestonice16', 'Belgium_UP_GoyetQ116_1', 'Russia_West_Siberia_HG', 'Serbia_IronGates_Mesolithic', 'Karitiana.DG', 'Papuan.DG', 'Armenia_LBA.SG', 'Iran_GanjDareh_N', 'Turkey_Epipaleolithic', 'Morocco_LN.SG', 'Cyprus_C', 'Russia_Boisman_MN', 'Romania_C_Bodrogkeresztur', 'Croatia_EIA', 'Netherlands_EIA', 'Russia_Samara_EBA_Yamnaya', 'Czech_CordedWare', 'Lithuania_EMN_Narva', 'Turkey_Arslantepe_LateC', 'Israel_C', 'Iraq_PPNA', 'Lebanon_ERoman.SG', 'ONG.SG')

Let me guess, you think any of the aggregate samples you use in any model are going to be the same? Outside of idential twins, even you and your brother will get a different genetic lottery despite sharing the same parents. Let alone a migration period aggregate. The point of using aggregates is to get a picture of the overall population.

Last but not least. publicly admitting to fitting your tail to prove your hypothesis?

Your tail is out of whack. If it's what the Serbian team produced, it is junk, 30+ full-time researchers were devoted to that. The so called scientists.
I based mine through trial and error of two long days.

Bet you were sweating. But do not misconstrue what I said. It is the labeling of the aggregates per the Roman Frontier paper supplement. Table 3 of the csv part of the supplement.
My tail removed stuff that were there for fluff, Spain IA etc, which did little as far as improving the fit. This way you leave a performing genral tail, which then you modify per experiment you run. You add a population that is variably related to the components you are testing. So for early-Albanian and migration-Slavic you add to the tail Albania_BA_IA and Czech_Early_Slav. This way the tail is designed to give you a better read on the admixture of the populations in your experiment. https://academic.oup.com/genetics/article/217/4/iyaa045/6070149

PS:

1)Here I just dropped the Netherlands_MBA_IA and Iberia_IA since for Balkan pops they are irrelevant historically but also add nothing to the tail, if anything they made the analysis worse.

2)I changed the CroatiaMLBA_SloveniaIA (aggregate constructed by the authors, God knows by what logic, but they even came up with worse~ CroatiaSerbia_RomanAnatolian, LOL) cause it represents a ghost population, an aggregate of different profiles that neither lived at the same time neither share a profile, instead added whatever was left of Croatia_Cetina.

3)I added Ukraine_EBA_Yamnaya and Suddan Early Christian (part of the pre print 2 years ago), because it dramatically improved the tail (Ukraine_EBA_Yamnaya less so, but still substantial).

That concludes the "general", all purpose tail. The final change is per the practices recommended by the Patterson paper on qpADM on how to construct a tail for a given experiment.

4)Thus, I added Czech_IA_Hallstatt and Czech_Early Slav as the population variably related to the Avar_Slavic component of the test compared to the Albanian Medieval component. And Albania_BA_IA and Bulgaria_IA as the population variably related to Albanian Medieval compared to the Avar_Slavic component of the test.

Compare:


right = c("OldAfrica", "Steppe_BA", "EHG", "Iron_Gates_HG", "Anatolia_N", "Ukraine_EBA_Yamnaya", "Sudan_EarlyChristian", "Kazakhstan_EarlySarmatian","Czech_EarlySlav_660-770",
"Iran_N", "Greece_Minoan","Croatia_MBA_Cetina", "Czech_IA_Hallstatt", "Bulgaria_IA","Albania_BA_IA",
"Steppe_IA","SoutheastTurkey_MLBA", "Baltic_BA")

 

[PaleoRevenge]

That was the rhetorical question I think I went out of my way to help you understand. Here to make it plain for you, you don't!

No you like to talk a lot and fancy yourself. Keep in mind I got a working model within two days of installation. Take note of that.

I am not sure what your models having a cline (all models do lol) have anything to do with this. And yes you can model Albanians with Illyrian + Slavic as did Lazaridis and even Olade and the Serbian academy did, whether you like it or not. Also enlighten me how is my model forced? I literally even removed all the unnecessary fluff from the Olade tail, the tail literally works with just 17 populations. This is the tail I had on hand and had to force nothing. Meanwhile to make your test work you had to adjust your tail for 2 days? With 24 populations, now that is forcing and overfitting a model my friend. You are going based on the result you "want" to find, and over fitting the model till it passes.

Forcing? I weighted my sources to produce circus models like below? Explain this to me.

Your tail is wrong bro, look at mine. P-value 90%+

Your tail also does this.

you want to debate this? LMAO

You did not?

I left Estonia_BA description, but I was hiding something? You're just being stupid.

But either way, how are Baltic BA better proxies for Slavs when not only they lived thousands of years ahead of the Slavs, but outside of being connected to each other due to Baltic BA which contributed to the Baltic drift in Slavs, with the additional steppe they had creating the Balto-Slavic artifact, then samples like Czech_EarlySlav (nomenclature by Reich himself) or Av2, which we know for a fact lived at the very early stages of Slavic (proper) ethogenesis and shared early Slavic cultural complexes?

Unmixed medieval Slav carry 59% Estonia_BA, a component most of the IA Balkans did not carry, so it can be used as a proxy to measure Slavic drift.

Also bible thumber, there were among the Viking samples, several pure Slavs, purer than Czech but less so than the recently released Polish samples. Unlike the Polish mdv samples, these guy are in the reich database.

Hmm, I wonder where we could find a population like that... Lacking any sort of Slavic, and rich in Albania_BA_IA + Eastern / Anatolian
Did you even read the models I shared with you? With your own "proxies" lol.

How is your model stretched to get the reconstruction I desire when you "literally" came up with this garbage?
If I was to do it, it would look something like this.

The same tail that produced the circus acrobats?

Let me guess, you think any of the aggregate samples you use in any model are going to be the same?

Aggregate two samples just because of modern borders, even if they show no relation to one another. You should not even have to combine them. If what you believe is true, you should model postmdv-kukes to kenete, they are from same exact location, kente is 850 AD, Kukes range from 1400 to 1700 AD. Feel free to demonstrate.

Bet you were sweating.

My tail removed stuff that were there for fluff, Spain IA etc, which did little as far as improving the fit.

You did what I did, except my tail does not make Italians Slavic and central Balkan IA. I started initially with a two way model, doing what you did and it led to a great fit but the results were inconsistent with other models. Than I tried again using a three-way model for Montengrin mdv Slavs as Cinamak, mdv-Slav, and MENA. I looked at Davidski's paper, also Danubian and borrowed proxies by trial and error, also borrowed two of yours. My model is consistent. Now if you want to test Italians and want to measure the Germanic drift, it probably requires of changing Iron Gates to Villanueva HG, Croatia EIA to Spain IA, Narva MN to Denmak LN. You have to be geographically weary of who you are measuring and the history of the region.

How did I stretch my model when you had to add 24 populations to your tail mate? Your tail has plain, .DG, .SG, single sample populations,

The single Papuans and Amazonians are preventing Illyrian heritage from being detected? LOL Oh no.
Maybe my single Armenian sample could be replaced with Armenian IA, maybe Croatia EIA can be replaced with Cetina or Bezdanjaca BA to improve fit, other than that, it's a fine model.

 

[PaleoRevenge]

Point here is while the Late Antiquity Naisus sample could be viable for all we know, but this is not the way to test it. Where would you find Satsurblia like unadmixed profile in Late Antiquity Balkans?

My earlier explanation is validated. I went back to the early days of testing various Avar period E-V13s and one fellow was particularly Bassarabi-like in profile but with MENA admixture. So is qpdam really saying you need pure CHG or that's the best way it can compensate when using pure Iron age profiles? Let's compare a Bassarabi profile that has MENA mixture built in. Let's see who is right.

compare to qpdam, note both samples are E-V13.

So what if I use a MENA admixed version of Hungary I18832_E-V13, will it eliminate the need for Satsurblia? hmmmmm

Well hot damn, both (I14686 and I14687) and I15707, pass. There goes your contemporary wining. The only issue with I16750, is that is has too much MENA so it's exaggerating Slavic. What if I combine I18832 with I16750 to produce a hypothetical ghost population of proto-Alb that carried less MENA?

Now the Slavic admixture is quite accurate.
And three kukes post-mdv combined as bonus.

Also another bonus.

 

[PaleoRevenge]

right = c('Cameroon_SMA', 'Czech_Vestonice16', 'Belgium_UP_GoyetQ116_1', 'Russia_West_Siberia_HG', 'Serbia_IronGates_Mesolithic', 'Karitiana.DG', 'Papuan.DG', 'Iran_GanjDareh_N', 'Turkey_Epipaleolithic', 'Cyprus_C', 'Russia_Boisman_MN', 'Romania_C_Bodrogkeresztur', 'Croatia_MLBA', 'Netherlands_EIA', 'Russia_Samara_EBA_Yamnaya', 'Czech_CordedWare', 'Lithuania_EMN_Narva', 'Turkey_Arslantepe_LateC', 'Israel_C', 'Iraq_PPNA', 'ONG.SG')

 

[Archetype0ne]

Very interesting right? If you use some sort of methodology the models themselves would lead you to something quite surprising... and historically hard to believe (per the few people I showed).

But I am afraid that point is beyond petty stuff like BG_IA or Albania_BA_IA models Albanians, or how much Slavic admixture Albanians have.

Try comparing contemporary Slavic sources for proto Albanians. Also check again, the two Alb_mdv samples are not from the 1600s, they are 770-960, right about the time of the first Kingdom of Arber...

You might have caught on to the "bonus:1%" in my earlier post. Hopefully consciously. But it seems you took the hint and ran completely the opposite direction...

Once I get some replies from pepople familiar with the history of the period, it might be worth sharing more. But so far it seems there is no historical record to speak of, for the weird conclusion the models lead to...

“What can be said at all can be said clearly, and whereof one cannot speak, thereof one must be silent.” Wittgenstein

 

[Jovialis]

I made this

 

[Archetype0ne]

I made this

Very viable graph Jov.

I would add that the motivation also matters a lot. 1st "peak" - I think I found out; 2nd "valley" - I want to prove my point; 3d "plateau" - What is even going on?

As we have seen time and time again, these two graphs do not always align 1 to 1. Whether it is because of institution affiliation, personal grudges, funding or what not, people at the plateau of confidence might be anywhere on the graph of motivation. I would argue maybe Lazaridis and Patterson are some of the few of the major players at the plateau on both graphs.

Take Olalde for example, the institution possibly needed to prove a point per directives, but the fact that they are professionals at the plateau of confidence gave us the fact that they had to label their graphs "Eastern Ancestry" despite the name of the paper. It's a very fine line.

When I reviewed the pre-print I think I was at the valley of both. "I started using qpADM" & "I want to prove a point". Maybe my convictions of Slavic % in Albanians lead the building of the models. Had very similar right tail build to Paleo, and selected the left tail similarly to prove a point. Result 12-20% Slavic, 12% for myself.

This time around I completely dropped the tail (constructed to prove my point), went with the Olalde tail and the Patterson guide to fine tune it, as well as I could. And selected the left populations based on events we know from history and contemporary sources. To my surprise the Slavic admixture in Albanians jumped to 24% and 33% for my own raw data. But I am no longer trying to prove a point, just want to find out. The models that follow after this point, by just manually testing on the left tail are what is surprising. Hopefully someone is familiar with Byzantine history to help me make sense of them. Posting now, as I compile them.

 

[Archetype0ne]

Spoiler: Previous Post

View attachment 14925

1)Here I just dropped the Netherlands_MBA_IA and Iberia_IA since for Balkan pops they are irrelevant historically but also add nothing to the tail, if anything they made the analysis worse.

2)I changed the CroatiaMLBA_SloveniaIA (aggregate constructed by the authors, God knows by what logic, but they even came up with worse~ CroatiaSerbia_RomanAnatolian, LOL) cause it represents a ghost population, an aggregate of different profiles that neither lived at the same time neither share a profile, instead added whatever was left of Croatia_Cetina.

3)I added Ukraine_EBA_Yamnaya and Suddan Early Christian (part of the pre print 2 years ago), because it dramatically improved the tail (Ukraine_EBA_Yamnaya less so, but still substantial).

That concludes the "general", all purpose tail. The final change is per the practices recommended by the Patterson paper on qpADM on how to construct a tail for a given experiment.

4)Thus, I added Czech_IA_Hallstatt and Czech_Early Slav as the population variably related to the Avar_Slavic component of the test compared to the Albanian Medieval component. And Albania_BA_IA and Bulgaria_IA as the population variably related to Albanian Medieval compared to the Avar_Slavic component of the test.

Compare:


right = c("OldAfrica", "Steppe_BA", "EHG", "Iron_Gates_HG", "Anatolia_N", "Ukraine_EBA_Yamnaya", "Sudan_EarlyChristian", "Kazakhstan_EarlySarmatian","Czech_EarlySlav_660-770",
"Iran_N", "Greece_Minoan","Croatia_MBA_Cetina", "Czech_IA_Hallstatt", "Bulgaria_IA","Albania_BA_IA",
"Steppe_IA","SoutheastTurkey_MLBA", "Baltic_BA")

Continuation of my previous post.

Goal of the test: Model modern Albanians taking into the account population movements of the Migration Period.

I take the previous tail and I test if Albania_BA_IA + Migration Period Slavic proxies work:

At this point I could have called it a day, messed with the right tail till I get a model compatible with my hypothesis, 10-15% Slavic in Albanians from my last review in Danubian Limes a year or two ago. But that is not the point, I want to actually find out.

So I test the other modern Albanian samples/aggregates:

Spoiler: Models 1

Well clearly the mode is not really working. Even if I could make it work for one of the samples, the likelihood of deriving any truth from it is low. So at this point I tested with the other Slavic proxies. What I was looking for most forward in this paper was the new Balkan Slavic Proxies we got. So I went ahead and filtered the Olalde et al table 6 of the supplement for the Slavic Samples(Balkans Iron Age+Anatolia Roman+Eastern Europe) between 661-1121 with at least 40% CEE ancestry. This surely ought to fix the problem.

Spoiler: Filtering

But the results were not as expected. Instead of a better proxy, these new samples seem to be a worse proxy!

Spoiler: Results with the new proxy

Well. Maybe the fact that the Balkan_Slav_661-1121 proxy performs worse than the Avar_Slavic_540-640 proxy has something to do with our use of Albania_BA_IA. What if we use the Albania_Medieval_773-989 proxy for the local ancestry? That surely will fix it. These are samples from right about the time Progon's first Kingdom of Arber is mentioned (Progon, Lord of Kruja). Between Albania_BA_IA to Albania_Medieval_773-989 no Slavic admixture seems to have happened since all models fail spectacularly no matter what proxy one uses, but some 1/3 admixture of something Anatolian or further Eastern has been ingested (this requires a separate experiment and a separate detailed post as it is outside our current scope). So lets look at the models with Albania_Medieval_773-989 + Balkan_Slav_661-1121.

Spoiler: Albania_Medieval_773-989 + Balkan_Slav_661-1121

Models as expected start working much better, hypothetically a better admixture ratio, as well as better standard errors than previously. Contemporary sources, all files on the tail same sequencing (no: noUDG, SG, DG etc). But the only problem is that the model still doesn't pass. Maybe the Balkan_Slav_661-1121 proxy is actually not the best proxy we have. Lets test again with the Avar_Slavic_540-640, which per Ph2ter (very knowledgeable of early Slavic history) is the best proxy we yet have.

Spoiler: "Albania_Medieval_773-989", "Avar_Slavic_540-640"

Much better proxy overall, can model multiple Albanian samples from Modern Era to current... And still none of the models pass.

 

[PaleoRevenge]

The best way to get a ballpark of Slavic admixture is to use phiter's Slavic drift calc.

BSD,0.139782,0.1522635,0.124916,0.0901845,0.1006165,0.0314615,0.0163085,0.0316,0.0371455,-0.004572,0.0012435,-0.0218345,0.039901,0.0662735,-0.0386175,-0.0082235,0.002497,0.0084965,0.00349,0.005984,0.009297,-0.003105,-0.0082725,-0.0721915,0.005406
EEF,0.116187154,0.182560923,0.003249077,-0.099981231,0.050991615,-0.048119308,-0.003850385,-0.007774846,0.040007923,0.082146538,0.008793923,0.011436077,-0.023797077,6.35385E-05,-0.044860692,-0.010321692,0.024121077,-0.000838154,0.011564154,-0.009398769,-0.015597385,0.006078,-0.004531692,-0.004096846,-0.005775538
WHG,0.130612,0.1112005,0.192897,0.20001875,0.14579625,0.056824,0.01562825,0.03432525,0.081912,-0.00350825,-0.01221975,-0.0175345,0.0237115,-0.00158275,0.047638,0.0614885,0.001532,0.0066195,-0.006128,0.054776,0.0923685,0.010696,-0.0403635,-0.1486655,0.01676475
EHG,0.122929,0.032496667,0.129980667,0.209843333,-0.010668667,0.056800333,-0.019036,-0.023921667,-0.002318,-0.086683333,0.017646,-0.018783333,0.033399,-0.038855667,0.018729,0.029258333,-0.014299,0.003040667,-0.003854667,0.012214,-0.007029333,0.015951,0.010558,-0.020886667,-0.009101
CHG,0.084229,0.1045995,-0.0914515,-0.003876,-0.089863,0.019662,0.0210335,-0.0036925,-0.129975,-0.076995,-0.0051965,0.02248,-0.049058,0.002133,0.029044,-0.0241975,0.029141,-0.0064615,-0.0189175,0.037768,0.0343145,-0.0069245,0.003266,-0.020846,-0.0006585
IRAN,0.045339667,0.065332333,-0.1523565,0.008021167,-0.1233565,0.026913,0.0137875,0.000846167,-0.081639,-0.056797,-0.003545333,-0.0004995,0.0043855,-0.008876833,0.0327765,0.054472333,-0.005346,0.005764333,0.013721833,-0.033578667,0.006883667,-0.029202667,-0.008771333,-0.0370135,0.0206965
SIBERIA-S,0.0665865,-0.221893,0.090886,0.088179,-0.0947865,-0.004602,-0.0300815,-0.031268,-0.0165665,-0.040274,0.0039785,-0.015586,0.0173185,-0.0526405,0.0051575,0.0027845,-0.004498,-0.007791,-0.001634,0.006816,-0.0295105,0.0115,0.01177,0.008977,-0.0222735
SIBERIA-W,0.0839445,-0.136081,0.0832495,0.15108425,-0.090863,0.02767975,-0.048882,-0.047998,-0.02684375,-0.06496725,0.0302855,-0.00588225,0.0239715,-0.0568725,0.01543825,0.0115355,-0.0144725,0.00272375,0.00399075,0.006128,-0.0354685,0.00995425,0.01996625,0.004097,-0.00805325
AMUR,0.019729333,-0.450895,0.079824,-0.055986667,-0.050060667,-0.045366333,0.013709333,0.010461,0.007294667,0.015004,-0.047092333,0.001698333,0.002329,0.017386667,-0.000724,-0.005303667,-0.005041667,0.000718,0.008966333,0.019967667,-0.00104,-0.026338333,-0.021280667,-0.003253333,-0.019958333
LEVANT,0.0685785,0.17289375,-0.027341,-0.14333225,0.028082,-0.064633,-0.01721475,-0.01407625,0.08094025,0.03603725,0.012504,-0.01618575,0.0353815,0.000688,-0.01347025,0.002917,0.00440025,-0.00456075,-0.006976,0.01572625,-0.00414875,0.005317,-0.002311,-0.00478975,-0.001946
MOROCCO,-0.189857,0.0812424,-0.0233816,-0.085918,0.026897,-0.0562244,-0.0688578,0.0189222,0.1556838,0.0023324,0.0228318,-0.0328806,0.0757278,-0.0494342,0.0694074,-0.035799,0.007719,-0.0649408,-0.1416618,0.0393438,-0.037908,-0.1254826,0.0707936,-0.0144358,0.0191596
CHINA,0.01935,-0.442263,0.0126335,-0.0560405,0.043854,0.0105975,0.003995,-0.000692,0.000102,0.0071075,-0.092724,-0.0129635,0.012859,-0.005849,-0.0065825,0.0005965,0.0022815,-0.00152,-0.002577,-0.005878,0.012041,0.01558,0.0156525,-0.000181,0.006227
INDOCHINA,-0.0034145,-0.334617,-0.0708985,-0.0040375,0.1024805,0.0292835,-0.00893,0.000577,0.015237,-0.006652,0.043601,0.007568,-0.0066895,-0.011216,-0.0050895,0.004243,0.005802,0.0020905,-0.0079815,0.0005,0.0043675,0.0072335,-0.002773,-0.0092785,0.0054485
ASIAN-PACIFIC,0.009675,-0.4087505,-0.0511,-0.0363375,0.1120205,0.040997,-0.00329,-0.00773,-0.009408,-0.0088385,0.0369435,0.002623,-0.0068385,0.0028905,0.007736,0.0082205,0.0024775,0.004434,0.0016345,-0.0116305,0.003743,-0.013911,0.0036975,-0.0134355,-0.0332905
INDIA,0.034147,-0.081242,-0.179887,0.122741,-0.096941,0.068886,0.00094,0.006923,0.035383,0.020228,-0.005034,0.004946,-0.006838,0.007019,0.002172,0.004375,0.007041,0.007348,-0.001383,0.00025,0.004367,-0.008779,-0.000863,-0.001566,-0.007065
SOUTHEAST-ASIA,-0.0381305,-0.216308,-0.1384035,0.1313,0.056626,-0.089942,-0.004935,0.004961,0.037837,0.016219,0.0107175,0.000075,-0.001561,0.0012385,-0.000407,-0.002188,0.0021515,0.0001265,-0.004462,0.0194465,0.0074245,0.0075425,-0.007703,0.0013255,-0.002455
WEST-AFRICA,-0.6038315,0.0573775,0.0139535,0.0272935,-0.00277,0.0027885,0.1236155,-0.0956495,0.0122715,-0.005285,-0.001299,-0.0219555,-0.021184,-0.0037845,0.0023075,0.0005965,0.011278,0.0184965,-0.003331,-0.001876,-0.0028075,0.000618,-0.000801,-0.00241,-0.0034725
EAST-AFRICA,-0.548058,0.051792,0.011125,0.0172805,0.001077,-0.001952,0.143709,-0.1073035,0.0579825,-0.037085,-0.003654,-0.0200075,-0.023563,-0.003303,0.0158795,-0.0101435,0.028489,0.1040115,-0.0267735,-0.0056275,-0.012166,0.0033385,0.001787,-0.004579,-0.0028735
SOUTH-AMERICA,0.050113583,-0.311090528,0.120835611,0.096469333,-0.111858417,-0.019909778,-0.288521306,-0.344466306,-0.010714917,-0.014796444,0.000320361,-0.002056472,-0.001238944,0.02309375,-0.007886694,0.004990528,0.005255028,-0.000355306,0.005474806,0.00577025,-0.001382944,0.004994139,0.000034361,-0.000796556,-0.001284111
CENTRAL-AMERICA,0.04467525,-0.317353,0.1119105,0.09197425,-0.11640625,-0.0169425,-0.2842455,-0.3302745,-0.00869225,-0.014761,-0.00024375,-0.0058075,0,0.02373975,-0.00257875,0.00477325,0.0079535,0.00072825,0.00339375,0.0005315,-0.0020275,0.012087,-0.00221875,-0.000512,-0.00302375
NORTH-AMERICA,0.047426333,-0.307367,0.115273333,0.090655333,-0.104327,-0.015431667,-0.254438333,-0.302141,-0.011658,-0.020228333,0.000541667,-0.00005,0.000099,0.013991667,-0.009862333,0.008486,0.005519333,0.003251667,0.004315333,0.004627667,0.003494,-0.002473,-0.003861667,0.002972,-0.000678667
CANADA,0.052928,-0.3234465,0.1071025,0.080427,-0.1029425,-0.0170125,-0.195294,-0.2371055,-0.005011,-0.016219,-0.004547,0.0020235,0.000223,-0.005161,-0.0077365,-0.003845,0.0009775,0.003674,0.006222,0.0113805,0.0024955,-0.0061825,0.0067785,0.0063865,0.0055085
AUSTRALIA,-0.042115,-0.224432,-0.205154,0.215442,0.145566,-0.322398,-0.007285,0.014538,-0.000818,0.003098,-0.004872,-0.002248,0.004311,-0.001239,-0.008415,-0.000796,0.004694,-0.001774,-0.01169,-0.003377,0.003369,-0.006801,-0.000863,-0.001928,-0.00946
YAMNAYA,0.125837778,0.089253889,0.042907889,0.115455556,-0.027868444,0.044684556,0.004491111,-0.002948667,-0.054857889,-0.072995667,0.001858333,0.000349667,-0.001651778,-0.023609889,0.037262889,0.015734,1.11111E-07,-0.001478,-0.001704,0.012505889,-0.003119667,0.001374,0.011229222,0.018436222,-0.004523667

When I ran 47 Albanians samples, the average was 10.7, pure mdv Slavs come out to 41.6. This implies 10.7/41.6 = 25.7% Slavic admixture. BUT we also have to look at the proto-Albanoid population.

I18832 has 2.0 Slavic drift built in. I would assume proto-Alb was 70-75% I18832, and 25-30% MENA shifted. That means about 1.4 of the Slavic drift was already built in with the early unadulterated Albs.
Also Albs are about 2.5% Germanic based on my G25 calc. That means an additional Slavic drift came from Goths (.025 x 6.2 = 0.155). So 0.155 + 1.4 = 1.555, this amount of Slavic drift is not due to Slavs and has to be subtracted from 10.7, which makes Slavic contribution to Albanian Slavic drift = 10.7 - 1.555 = 9.145, that yields a Slavic admixture of 9.145/41.6 = 21.98%.

Using the tail I posted two days ago, the best proxy that gives similar Slavic admixture is to combine I18832 with I16750 as a proxy for the proto-Alb ghost population of 500 AD.

The same tail, also produces this.

That is impressive that it could make three-way model for medieval Slavs of Montengro and correctly picks up a Illyrian based substrate.

 

[PaleoRevenge]

Try comparing contemporary Slavic sources for proto Albanians. Also check again, the two Alb_mdv samples are not from the 1600s, they are 770-960, right about the time of the first Kingdom of Arber...

Kenete is 850 AD. He is also T2a which is indication that he could be a foreigner that later became Albanian. Post mdv are 1400-1700 AD. So if Kenete was an Arber, you should be able to model post-mdv Kukes with kenete, location wise they are only 6 miles apart.. Which does not work at all.

Also I could model all Albanian samples with Avar_E-V13 plus pure mdv Slav, but not Kenete.

The Avar E-V13 that works so well, is likely not a local of Hungary but one of many of the resettled captives the Avars picked up on their routine raids in Serbia.

 

[Archetype0ne]

Continuation of my previous post.

....


Models as expected start working much better, hypothetically a better ratio, as well as better standard errors than previously. Contemporary sources, all files on the tail same sequencing (no: noUDG, SG, DG etc). But the only problem is that they still do not work. Maybe the Balkan_Slav_661-1121 proxy is actually not the best proxy we have. Lets test again with the Avar_Slavic_540-640, which per Ph2ter (very knolwedgable of early Slavic history) is the best proxy we yet have.

Spoiler: "Albania_Medieval_773-989", "Avar_Slavic_540-640"

Much better proxy overall, can model multiple Albanian proxies from Modern Era to current... But still none of the models pass.

At this point it was hard to resist the urge to mess with the right tail. But I did. So lets keep following what the models are telling us. If Avar_Slavic_540-640 is a better proxy than Balkan_Slav_661-1121 why would that be? What if we add to the model an Avar proxy. Most of the time (99% from my experience) if you add a third component to an experiment that fails with two it will fail even more spectacularly. But hey, worth a shot at this point.

Spoiler: "Albania_Medieval_773-989", "Avar_Slavic_540-640", "Hungary_EarlyAvar"

It did not only perform better, but the standard errors are better across the various modern Albanian profiles. But it still failed. Maybe we need to retry the Balkan_Slav_661-1121 proxy with the Hungary_EarlyAvar.

Spoiler: "Albania_Medieval_773-989", "Balkan_Slav_661-1121", "Hungary_EarlyAvar"

What?! It still performs worse than Avar_Slavic_540-640 even with the Hungary_EarlyAvar there to compensate.
At this point a week or two ago, I was tired and discouraged. But before hitting bed I tried the last run. And this left my jaw hanging. Why? How?

Spoiler: "Albania_Medieval_773-989", "Hungary_EarlyAvar"

I was shocked to say the least. I went on to try with my personal raw data, and various other Albanians. The model outperformed all other models for everyone. But for some Albanians, including me (South Albanian mothers side, from a place with known Slavic settlement) and a friend (North Albania bordering Montenegro) who needed a Slavic source as well. I have had no time to test further during the week so this is where I stopped.

Spoiler: A1

Since I ran this I got in touch with a few people with a History background. They were as surprised as me. We know the Avars were big players during the period between 600 till their demise in the late 800s. They did have connections to cities like Timacum Minus, Viminacium, even Singidinum. But we are not aware of them ever settling in Albania. But then again this a period of Byzantine history where sources are very scarce. Although Komani-Kruja seems to have had attested trading links to Hungary:

Objects from a vast area covering nearby regions the entire Byzantine Empire, the northern Balkans and Hungary and sea routes from Sicily to Crimea were found in Dalmace and other sites coming from many different production centres: local, Byzantine, Sicilian, Avar-Slavic, Hungarian, Crimean and even possibly Merovingian and Carolingian.

The question is, did this pots come with people.

 

[Johane Derite]

Since I ran this I got in touch with a few people with a History background. They were as surprised as me. We know the Avars were big players during the period between 600 till their demise in the late 800s. They did have connections to cities like Timacum Minus, Viminacium, even Singidinum. But we are not aware of them ever settling in Albania. But then again this a period of Byzantine history where sources are very scarce. Although Komani-Kruja seems to have had attested trading links to Hungary:

The question is, did this pots come with people.

Plenty of scholars have connected the treasures of Vrap and Ersekë in Albania to the Avars. Some even argued the proto-Albanians came from the Sermesians led by Kuber.

 

[Archetype0ne]

Plenty of scholars have connected the treasures of Vrap and Ersekë in Albania to the Avars. Some even argued the proto-Albanians came from the Sermesians led by Kuber.

Interesting. In fact I recalled your quote a couple of years ago about "asiatic" skulls in Komani, when I saw these results. But the interesting thing is that with the little time I had to further test during the week, it was very hard to model this 13% Avaric component in Albanians with any Slavic. And even when I tried modeling Hungary_EarlyAvar itself with Slavic component in it, or better yet, with *anything* the models failed. Ph2ter told me that these Avars were pretty much East Asian, IA or EMA Mongolian like per his G25.


Ph2ter:

The best proxy for Slavic I think still is Av2 sample.

Early Avars came from the Far East. Autosomally they are Mongolian.

And while I could confirm his first statement. His second statement I was unable to confirm (lack of time, did not try many models, plus my knowledge of steppe history is too weak to construct proper qpADM models).

Based on the models currently this

Plenty of scholars have connected the treasures of Vrap and Ersekë in Albania to the Avars.

makes a lot of sense.

This not so much:

Some even argued the proto-Albanians came from the Sermesians led by Kuber.

But overall this *could* imply that the Slavic / Eastern admixture in Albanians did not generally or initially admix into early Albanians from a Balkan_Slav_661-1121 like profile, but rather an Avar profile.

Later on though, as is the case with my raw data, and some others I have access to, for some both are required. But from the runs I did this was only the case for Albanians deriving ancestry from South-East Albania and regions bordering Montenegro, with are known to have had Slavic settlements/toponyms. While Hungary_EarlyAvar was the case for everyone.

 

[Johane Derite]

Interesting. In fact I recalled your quote a couple of years ago about "asiatic" skulls in Komani, when I saw these results. But the interesting thing is that with the little time I had to further test during the week, it was very hard to model this 13% Avaric component in Albanians with any Slavic. And even when I tried modeling Hungary_EarlyAvar itself with Slavic component in it, or better yet, with *anything* the models failed. Ph2ter told me that these Avars were pretty much East Asian, IA or EMA Mongolian like per his G25.


Ph2ter:



And while I could confirm his first statement. His second statement I was unable to confirm (lack of time, did not try many models, plus my knowledge of steppe history is too weak to construct proper qpADM models).

Based on the models currently this

makes a lot of sense.

This not so much:



But overall this *could* imply that the Slavic / Eastern admixture in Albanians did not generally or initially admix into early Albanians from a Balkan_Slav_661-1121 like profile, but rather an Avar profile.

Later on though, as is the case with my raw data, and some others I have access to, for some both are required. But from the runs I did this was only the case for Albanians deriving ancestry from South-East Albania and regions bordering Montenegro, with are known to have had Slavic settlements/toponyms. While Hungary_EarlyAvar was the case for everyone.

Yes the asiatic skulls in koman were classified as avars by Bowden and Buchet, as they had avar goods.

 

[Archetype0ne]

Yes the asiatic skulls in koman were classified as avars by Bowden and Buchet, as they had avar goods.

Was it 3/25 skulls or am I misremembering? Cause coincidentally 3/25 = .12... Which is also the ratio we get for Avaric in the models lol. But this is just fun trivia. Doubt anything can be construed.

Further fun speculation:

Kuber[3] (also Kouber or Kuver) was a Bulgar leader who, according to the Miracles of Saint Demetrius, liberated a mixed Bulgar and Byzantine Christian population in the 670s, whose ancestors had been transferred from the Eastern Roman Empire to the Syrmia region in Pannonia by the Avars 60 years earlier.[4][5] According to a scholarly theory, he was a son of Kubrat, brother of Khan Asparukh and member of the Dulo clan.

Kuber

All those L283, z2103 and V13 with south Balkan profiles in the Hungary paper in Pannonia would fit this like a piece in a puzzle.

I wonder what profile these "Bulgarians" had.
How East Asian, or Slavic it would have been?

 

[Archetype0ne]

Spoiler: Modeling Bulgaria_Medieval_oEastAsia

^fails standard error and p

^fails se

^fails p

^fails se

^fails se

^passes

Not sure if by medieval period the distances between West and East southern Balkans would have been homogenized. But these Eastern Asian outlier in Medieval Bulgaria seems to need a higher level of Yamnaya like ancestry to the level of Medieval Albania to be modeled. And it comes out as 60% Balkan profile (Medieval_Albania_773-989 proxy) + 40% Mongolian (Mongolia_Arkhangai_Xiongnu).

Spoiler: East Asian Proxies

Since the Xiongnu works almost as good as the earlyAvar proxy, I assume the reason I could not model the earlyAvar as any combination last time I tried is because as ph2ter says they are IA/mdv Mongolian with little other addition.

 

[PaleoRevenge]

I looked into the Bulgarian model problem. I tried many directions until it dawned on me, what the G25 is hinting is correct.

The mdv Bulgarian has Slavic-Illyrian mix component. This on the surface seems laughable but in the 800s, the Bulgars ruled almost half of the Balkans, including half of Hungary, where many Balkan natives were resettled by the Avars.

Let me demonstrate.

The model almost works with Kenete alone. With Shtike it is not even close. You know who else carries the Slavic to Illyrian component in 2 to 1 ratio as shown in the G25? The medieval Docleans.

I think it should become obvious by now that Kenete is half Kruja Komani, half Byzantine from Turkey. And the only reason he adds value into modeling Albanians is because of his partial Illyrian and partial MENA component which Albanians carry a minor substrate of both.

Also I tested out Av2, my Slavic proxy is noticeably better, models have tighter se and better p-value.

 

[PaleoRevenge]

I didn't realize the last model failed. Here is a better version, it fails but just like kenete it almost works, for the same reason I described.

 

[Archetype0ne]

I looked into the Bulgarian model problem. I tried many directions until it dawned on me, what the G25 is hinting is correct.

The mdv Bulgarian has Slavic-Illyrian mix component. This on the surface seems laughable but in the 800s, the Bulgars ruled almost half of the Balkans, including half of Hungary, where many Balkan natives were resettled by the Avars.

Let me demonstrate.

The model almost works with Kenete alone. With Shtike it is not even close. You know who else carries the Slavic to Illyrian component in 2 to 1 ratio as shown in the G25? The medieval Docleans.

I think it should become obvious by now that Kenete is half Kruja Komani, half Byzantine from Turkey. And the only reason he adds value into modeling Albanians is because of his partial Illyrian and partial MENA component which Albanians carry a minor substrate of both.

Also I tested out Av2, my Slavic proxy is noticeably better, models have tighter se and better p-value.

Glad you caught your error. Overall similar issues to earlier. You can get it to work for one pop but it will fail for other related pops.
With the current setup I got it to to pass with East Balkan source in a neat model, but a friend wants it posted on his blog, so decided not to share. The setup holds well for the populations I have in my dataset, from Bulgarians, to Croatians and Romanians.