"Ancient DNA reveals the origins of the Albanians" paper

[D.L.M]

Wanted to share how the same medevial Avar E-V13 sample that has the Bassarabi derived profile, holds up with the rest of the Albanian samples.

Personally I think the sample is a little extra Aegean shifted and it exaggerates Slavic admixture in profiles that don't have much MENA like kukes post-mdv. Even in Shtike, the Slavic admixture is a little inflated.

I did decide to experiment by combining Hun LaTene I18832 and the Avar I16750 as one by renaming both samples to have the same name to create a ghost population of 500 AD Albs that carries less MENA than I16750. This ghost profile would represent the "Dardanian cluster" as referred by rrenjet.

It fails for the modern Alb(south Albanian I think) because his MENA admixture is too high and needs a Bassarabi like profile with more MENA, which is why I16750 works well directly for this samples.

These models do not work at all for the Kenete sample, the closest I got to Kenete were these.

How much Illyrian ancestry do Albanians have according to this model?
And are R1b-Z2103 of Illyrian stock in Albanians?

 

[TetaEglantina]

How much Illyrian ancestry do Albanians have according to this model?
And are R1b-Z2103 of Illyrian stock in Albanians?

Maybe it is possible that r-z2103 was Anatolian, how else do we explain that crazy 43.8% Roman Anatolian input. I still think z2103 were proto Albanian speakers

 

[PaleoRevenge]

How much Illyrian ancestry do Albanians have according to this model?
And are R1b-Z2103 of Illyrian stock in Albanians?

R-Z2705 came with E-V13. I think R-Z2705 are Paeoni, like the laia(lala) tribe. Albanian Lala coincidentally are R-Z2705 FYI. What did the Paeoni profile looked like? Probably like BA Serbia samples but with a little more farmer and less WHG. They were initially similar to J2b, and had long contacts throughout the Bronze Age, in Iron Age, they fall under Thracian sphere. The way to truly know is to have samples of them.
Illyrian ancestry is likely similar to J2b-L283 ratio.
I did test out a model to construct the Arber population by merging I18832(E-V13), I16750(E-V13) and Doclea R3481(J2b-L283). The ratios are not entirely correct, as they should be 3 to 1 not 2 to 1, but that's all there is, in the database.
The model works, but not as good. For Kenete it never works, because simply it is not an Albanoid profile.

 

[PaleoRevenge]

Maybe it is possible that r-z2103 was Anatolian, how else do we explain that crazy 43.8% Roman Anatolian input. I still think z2103 were proto Albanian speakers

The study you are referring to, their model did not even pass, it's a fail. So why even bring up such non-sense model. If Albanians are 44% MENA, they would not be speaking Albanian. Not to mention that percentage is at odds with Albanian haplogroup portfolio.

 

[TetaEglantina]

The study you are referring to, their model did not even pass, it's a fail. So why even bring up such non-sense model. If Albanians are 44% MENA, they would not be speaking Albanian. Not to mention that percentage is at odds with Albanian haplogroup portfolio.

Brought it up because it's the latest study but it doesn't seem right. Did they explain if the CroatiaSerbia Roman anatolian samples were 100% anatolian, weren't they mixed with locals?

For example if they were 50% anatolian, that makes Albanians 21.9% Roman anatolian and how much of that is Greek instead?

25.2% South Illyrian
~ 21.9% Dalmatian Illyrian and Dacian
~ 21.9% Roman anatolian
31% Slavic and other central/eastern European

 

[PaleoRevenge]

Brought it up because it's the latest study but it doesn't seem right. Did they explain if the CroatiaSerbia Roman anatolian samples were 100% anatolian, weren't they mixed with locals?

For example if they were 50% anatolian, that makes Albanians 21.9% Roman anatolian and how much of that is Greek instead?

25.2% South Illyrian
~ 21.9% Dalmatian Illyrian and Dacian
~ 21.9% Roman anatolian
31% Slavic and other central/eastern European

That's correct, just because it got published does not mean it's blessed by God. The supplementary information to this paper is no longer public, you need a subscription to see it. I would like to see the tail they used for their model, just to test out my variables in their tail as a comparison. Also all their samples are not in reich databse yet, so we cannot test these new profiles. I can tell you on G25, none of these new IA profiles are good at modeling Albanians.

 

[blevins13]

Maybe it is possible that r-z2103 was Anatolian, how else do we explain that crazy 43.8% Roman Anatolian input. I still think z2103 were proto Albanian speakers

If Albanian is close to Messapic than E-V13 is out of the picture as proto-Albanian speakers.

 

[mount123]

The study you are referring to, their model did not even pass, it's a fail. So why even bring up such non-sense model. If Albanians are 44% MENA, they would not be speaking Albanian. Not to mention that percentage is at odds with Albanian haplogroup portfolio.

There's a Serbian crackpot with multiple sock puppet accounts in that other thread using this laughable piece of garbage model to claim Serbs have as much Paleo-Balkan DNA as Albanians. As in claiming Albanians are Middleeastern + Paleo-Balkan.

Indeed, the average Albanian haplogroup portfolio that is 80+ % Paleo-Balkan.

 

[PaleoRevenge]

There's a Serbian crackpot with multiple sock puppet accounts in that other thread using this laughable piece of garbage model to claim Serbs have as much Paleo-Balkan DNA as Albanians. As in claiming Albanians are Middleeastern + Paleo-Balkan.

Indeed, the average Albanian haplogroup portfolio that is 80+ % Paleo-Balkan.

The model se values fail. If you look at the excel, only the Serbian model passes, the rest either the p-value fails(not above .05) or se value fails (not below .05) and in some cases both criteria fail. The paper is comical, the alleged Serbian coding prowess turns out to be a paper tiger.
The same thing goes for their one way Iron Age models, half of them fail. What the hell were they doing for three-four years, the paper took at least a year to publish than got stuck over 2 years in pre-print.

Seriously, I produced a way better model(and they actually pass, convincingly too) for Bulgarians and Romanians within hours of trying. Who are these people?

Notice the difference between .DG and .HO samples. Having played with the larger dataset, I can tell you .HO samples are harder to model, because they are of inferior quality. The only reason their model barely passed for Serbians, it is because they added new samples for Serbs, they took some 20 new samples and sequenced them with the latest method, so they are comparing apple to apples. The Albanian .HO samples are hard to model, the quality is bad. The newer samples (southern arc) produce much more convincing results, for modeling purposes, I suggest people stick to .DG samples or generate your own like archetype did with the help of eupator.

 

[eupator]

The .DG samples have double the SNP coverage compared to .HO (1240K compared to 600K if I remember correctly).

 

[Moses 26]

The model se values fail. If you look at the excel, only the Serbian model passes, the rest either the p-value fails(not above .05) or se value fails (not below .05) and in some cases both criteria fail. The paper is comical, the alleged Serbian coding prowess turns out to be a paper tiger.
The same thing goes for their one way Iron Age models, half of them fail. What the hell were they doing for three-four years, the paper took at least a year to publish than got stuck over 2 years in pre-print.

Seriously, I produced a way better model(and they actually pass, convincingly too) for Bulgarians and Romanians within hours of trying. Who are these people?

Notice the difference between .DG and .HO samples. Having played with the larger dataset, I can tell you .HO samples are harder to model, because they are of inferior quality. The only reason their model barely passed for Serbians, it is because they added new samples for Serbs, they took some 20 new samples and sequenced them with the latest method, so they are comparing apple to apples. The Albanian .HO samples are hard to model, the quality is bad. The newer samples (southern arc) produce much more convincing results, for modeling purposes, I suggest people stick to .DG samples or generate your own like archetype did with the help of eupator.

What about Croats ?

 

[PaleoRevenge]

What about Croats ?

Croat samples are entirely .HO, which I prefer to not to spend time on. But for this one request, here you go, this model worked. P-value = 0.309, much better than the recent paper with 0.028 p-value.

Doclea J2b from Roman period is Illyrian with some MENA shift. I am sure other Roman era samples that are Illyrian derived would work as well too. I could not make a model work with IA Illyrian plus imperial Levant profiles and it's not worth spending time with .HO sample.

 

[PaleoRevenge]

Croat samples are entirely .HO, which I prefer to not to spend time on. But for this one request, here you go, this model worked. P-value = 0.309, much better than the recent paper with 0.028 p-value.

Doclea J2b from Roman period is Illyrian with some MENA shift. I am sure other Roman era samples that are Illyrian derived would work as well too. I could not make a model work with IA Illyrian plus imperial Levant profiles and it's not worth spending time with .HO sample.

Decided to do more runs. There were some other Coastal Croatia samples that also worked, but I think Doclea was the best p-value. I decided to run the Serbian Roman samples and surprisingly they work too, though the Celto-Illyrians from Slavonia make the best model. I would say Croats have a substrate from both, Illyrians and the nearby E-V13 populations of eastern Serbia, but the Illyrian substrate seem to be stronger.

 

[mount123]

Decided to do more runs. There were some other Coastal Croatia samples that also worked, but I think Doclea was the best p-value. I decided to run the Serbian Roman samples and surprisingly they work too, though the Celto-Illyrians from Slavonia make the best model. I would say Croats have a substrate from both, Illyrians and the nearby E-V13 populations of eastern Serbia, but the Illyrian substrate seem to be stronger.

Illyrian-like non-admixture auDNA in this context is a more fitting term. Both Illyrian Y-DNA and MtDNA are miniscule in Croats. The non-Slavic auDNA is probably something from eastern Tumulus Culture derived pops e.g. eastern-central European Celts aswell as some Western extensions of Basarabi Culture, so generally UF-like auDNA. This would be well in line with both Y-DNA and MtDNA distribution.

 

[TetaEglantina]

On Rrenjet I counted 240/1618 ydna that potentially could have been been MENA which is almost 15%

Didn't include any i2, i1, r1b (except for a couple which look non European), r1a, j2b l283, e-v13. In order to reach 40% MENA or even close to it some of these lines have to be Roman anatolian origin and the only 2 that are even possible are r-z2103 and r-pf7563 but the latter seems to be ancient greek according to bronze age samples and was already present in iron age Albania. The other possibility is the R-l51 samples which may be Roman in origin but mixed with Anatolian mtdna/women - there aren't that many R-l51 samples anyway to make it 40%

 

[PaleoRevenge]

On Rrenjet I counted 240/1618 ydna that potentially could have been been MENA which is almost 15%

Didn't include any i2, i1, r1b (except for a couple which look non European), r1a, j2b l283, e-v13. In order to reach 40% MENA or even close to it some of these lines have to be Roman anatolian origin and the only 2 that are even possible are r-z2103 and r-pf7563 but the latter seems to be ancient greek according to bronze age samples and was already present in iron age Albania. The other possibility is the R-l51 samples which may be Roman in origin but mixed with Anatolian mtdna/women - there aren't that many R-l51 samples anyway to make it 40%

Google R-Z2705. There is not a single ME individual on the family tree that is upstream. All ancestral clades are in Europe. J1 and T2 are MENA. Maybe some of J2a could be Anatolian, but likely most are local. There is not much ME paternal lines. What is quite ironic from the southern arc paper, the kukes Ottoman samples are low on Slavic and low on MENA, while the Gorani woman is high on slavic but also high on MENA. This suggest, ME admixture came by mixing with Balkan Slavs, these Slavs would have been similar to those of north Macedonia.

 

[TetaEglantina]

Google R-Z2705. There is not a single ME individual on the family tree that is upstream. All ancestral clades are in Europe. J1 and T2 are MENA. Maybe some of J2a could be Anatolian, but likely most are local. There is not much ME paternal lines. What is quite ironic from the southern arc paper, the kukes Ottoman samples are low on Slavic and low on MENA, while the Gorani woman is high on slavic but also high on MENA. This suggest, ME admixture came by mixing with Balkan Slavs, these Slavs would have been similar to those of north Macedonia.

But the study is suggesting Albanians have much more MENA than south Slavs? Also south Slavs have less potential MENA ydna lines

 

[PaleoRevenge]

But the study is suggesting Albanians have much more MENA than south Slavs? Also south Slavs have less potential MENA ydna lines

Serbs have less MENA, north Macedonians have more than Albanians.

 

[PaleoRevenge]

Ottoman Serbia samples from Sirmium. I don't think around 1500 AD Sirmium was Serbian, the area I think was Croatian. They model almost the same as Croatians but with more Slavic.

 

[mount123]

But the study is suggesting Albanians have much more MENA than south Slavs? Also south Slavs have less potential MENA ydna lines

That the models of this "study" are major fails was already pointed out, so why insist on it being right?

As for the potential Imperial era MENA-related patrilineage, not sure how you get to the 15 percentile as that's definitely not the case for Albanians anywhere. Funny enough, I just checked the modern Serbian samples (n=37) of that "study" and 24,324 % (n=9) carry Imperial era MENA lineages that first appear in the Balkans during the CE and are acompanied by a stark MENA auDNA profile. These include E1b-M81, J-M205, J1a-Z2215+, T1a-11151, Q-L245 etc.