Dodecad k12b West & Central Asian results Vol 4.

[Sterling Archer]

What is interesting about Astara is that the Iranian part is mainly Turkish-speaking, while the Republic part is Talysh speaking.

For some reason, I thought Iranian Astara was mostly Talysh )) Did Azerbaijanis historically dominate there or was there a recent migration to Astara, Iran?
Nevertheless, Gilan province overall seems to be predominantly Gilak and Talysh, so I assume the republic's Talysh population is related to them.

[Kyp]

For some reason, I thought Iranian Astara was mostly Talysh )) Did Azerbaijanis historically dominate there or was there a recent migration to Astara, Iran?
Nevertheless, Gilan province overall seems to be predominantly Gilak and Talysh, so I assume the republic's Talysh population is related to them.

I don't know why tbh. I think they are newcomers from Moghan and Ardabil mostly who settled down. At least this is what Iranian wiki claims. Also some turkish speaking Gypsies apparently.

Yes Gilan is almost completely Gilaki and Talysh (expept Astara)

[Sterling Archer]

This is not related to the topic, but I wanted to ask.
What is the difference between the calculators?

For example, dodecad k12 that we are sharing, and say PuntDNAL 13? Is one of them inherently more precise than the other? If so, then why?

I was running some results in PuntDNAL 13 for Azerbaijanis (Iranian, republic + Dagestan), and results while different than k12, still accentuates caucasian + turkic + iranic mix in Azerbaijanis. So rather logical.
Though for some reason, Caucasian element is linked with Ossetians + Abkhaz rather than Dagestani that Azerbaijanis get in dodecad k12.

[Leto]

This is not related to the topic, but I wanted to ask.
What is the difference between the calculators?

For example, dodecad k12 that we are sharing, and say PuntDNAL 13? Is one of them inherently more precise than the other? If so, then why?

I was running some results in PuntDNAL 13 for Azerbaijanis (Iranian, republic + Dagestan), and results while different than k12, still accentuates caucasian + turkic + iranic mix in Azerbaijanis. So rather logical.
Though for some reason, Caucasian element is linked with Ossetians + Abkhaz rather than Dagestani that Azerbaijanis get in dodecad k12.

There is no genuinely universal calculator for everyone. Some people get more accurate results from one calculator, others from another. The original Dodecad spreadsheet is horrible and shouldn't be used at all. I can't say much about punt, I think it's generally not bad, however no one has tried to revise or update it.

[Dr_Maul]

Is one of them inherently more precise than the other? If so, then why?

Most likely a combination of references and components. Dodecad components make much more sense in West Asia having both Gedrosia and Caucasus, compared to just "west asian" in K13. As well as Atlantic_Med and N Euro compared to "SW_Euro" (idk what this even is to be honest). K13 also gives me over 4% Mongoloid with oracles such as 95% Iranian + 5% Yakut which is definitely not accurate considering I barely score any in most calculators. (just a personal observation). As I said for references, K12b updated on Vahaduo has multiple regional Iranian averages with Arabs and Turkmens included, compare to just 'Iranians' on K13. These are just why its more useful for this region, for other areas it might be more or less accurate

[Sterling Archer]

Most likely a combination of references and components. Dodecad components make much more sense in West Asia having both Gedrosia and Caucasus, compared to just "west asian" in K13. As well as Atlantic_Med and N Euro compared to "SW_Euro" (idk what this even is to be honest). K13 also gives me over 4% Mongoloid with oracles such as 95% Iranian + 5% Yakut which is definitely not accurate considering I barely score any in most calculators. (just a personal observation). As I said for references, K12b updated on Vahaduo has multiple regional Iranian averages with Arabs and Turkmens included, compare to just 'Iranians' on K13. These are just why its more useful for this region, for other areas it might be more or less accurate

I see, makes sense. Although I think too much detalization might hurt the model, even though it might match someones expectations. Cue Gedrosia and Caucasus that seems to go hand in hand in Caucasus, and maybe don't need to be separated in some cases.
In essence, one might run into an issue of overfitting. Although, I don't no anything about the algorithms underlying the calculators, so maybe it is not an issue.

[Dr_Maul]

I see, makes sense. Although I think too much detalization might hurt the model, even though it might match someones expectations. Cue Gedrosia and Caucasus that seems to go hand in hand in Caucasus, and maybe don't need to be separated in some cases.
In essence, one might run into an issue of overfitting. Although, I don't no anything about the algorithms underlying the calculators, so maybe it is not an issue.

I think that component is actually why K12b is more accurate, because people have varying amounts. For example compare Abkhazians to Azeris in this example
Abkhazian has 19.42% Gedrosia, while Azeri has 20.81%. So, basically the same. But then where Abkhazian has nearly 56% Caucasus, Azeri has 38.42%. So if they were combined into West Asian, it would be kind of confusing or innacurate. Now for another, Turkmen_Uzbekistan has roughly the same 20% Gedrosia as the other 2. However it only has 19.93% Caucasus, compared to Azeri 38 and Abkhazian 55. So you can see why it can be useful for it to be split, compared to just "west asian".

[Sterling Archer]

I think that component is actually why K12b is more accurate, because people have varying amounts. For example compare Abkhazians to Azeris in this example
Abkhazian has 19.42% Gedrosia, while Azeri has 20.81%. So, basically the same. But then where Abkhazian has nearly 56% Caucasus, Azeri has 38.42%. So if they were combined into West Asian, it would be kind of confusing or innacurate. Now for another, Turkmen_Uzbekistan has roughly the same 20% Gedrosia as the other 2. However it only has 19.93% Caucasus, compared to Azeri 38 and Abkhazian 55. So you can see why it can be useful for it to be split, compared to just "west asian".

I agree with your point in the case of Caucasus.(Or we might add them together and call it CauCasia)
Merging NE and Atlantic_med though seems to be appropriate in the case of Abkhaz and Azeris for example (although Azeris seem to have much higher Atlantic_Med while NE is slightly less than Abkhazs), which leads me to my next point.

Which components and to what extent do we separate, while lumping together other ones? It seems inevitable to me that in the process some components, that should not be merged, will have to be arbitrarily grouped together with the assumption being that they are not that important for the region.
Depending how the equations work this might substantially affect the numerical results. Is there a scientific work regarding this topic?

[Zoro]

I agree with your point in the case of Caucasus.(Or we might add them together and call it CauCasia)
Merging NE and Atlantic_med though seems to be appropriate in the case of Abkhaz and Azeris for example (although Azeris seem to have much higher Atlantic_Med while NE is slightly less than Abkhazs), which leads me to my next point.

Which components and to what extent do we separate, while lumping together other ones? It seems inevitable to me that in the process some components, that should not be merged, will have to be arbitrarily grouped together with the assumption being that they are not that important for the region.
Depending how the equations work this might substantially affect the numerical results. Is there a scientific work regarding this topic?

With any admixture calculator based on modern human references the results don't literally mean you have ancestry from those references.

For example if you get 25% Gedrosian it doesn't mean you have 25% Baloch ancestry
If a Turk or an Arab has 12% Gedrosian it doesn't mean they have 12% Baloch ancestry
If a Bulgarian has 7% Gedrosian it doesn't mean Bulgarians have 7% Baloch ancestry

It does mean that you share 25% something (recent SNPs etc ) with Baloch and Bulgarians share 7% something with Baloch but it's up to you to figure out how come you share 25% something with Baloch. In other words it's up to you to figure out which ancestors, Iran-Chl or BMAC or whatever contributed this shared DNA between you and Baloch.

[Zoro]

The concept of one calculator fits all is flawed from the outset. There should be regional calculators based on ancestors relevant to that particular population since the Bronze Age or Chalcolithic.

The only scenario one calculator could be used by both Asians and Europeans is if it only has say 3 or 4 general components such as W Eurasian E Eurasian S Eurasian SSA that are relevant components for Asians and Europeans OR

if the calculator is based on really ancient populations such as WHG, ENF, ANE, AEE etc because both Asians and Europeans are descended from these few ancients.