I have other Southern European Admixture too?

[Chris596]

Alright, so I've noticed this a long time ago, but I can't explain it. I know that I'm already very Southern European (Southeastern), but I keep scoring smaller/bigger amounts of proper Southern European (like Sardinian/Italian/Spanish). Using my original Myheritage data and coordinates from GEDmatch, it appears on many different calculators. Also in Admixture Studio I score some Southern European.

If all of this is not enough, I also keep scoring some S. European on G25.

I have a confession to make: I know this is hilarious but when I posted my Admixture Studio results and some other results too, I always deleted the Southern European populations like a maniac, thinking that it can't be true so it won't appear in my results... So check this out:


What could be the reason behind it? Weirdly enough, I got Italian on FTDNA too


Vahaduo GEDmatch

Eurogenes EUtest
Target: Chris
Distance: 2.5116% / 2.51155634 | ADC: 0.25x
44.2 Serbian
35.4 Romanian
11.2 Hungarian
9.2 North_Italian

Eurogenes Jtest
Target: Chris
Distance: 2.3425% / 2.34247159 | ADC: 0.25x
47.8 HU
44.0 RO
5.2 North_Italian
2.4 AJ
0.6 Sardinian

HarappaWorld
Target: Chris
Distance: 1.2399% / 1.23985226 | ADC: 0.25x
80.2 Serbian
12.0 Italian
5.2 Romanian
2.0 Bulgarian
0.6 Japanese2

Dodecad K12b (Admixture Studio)
62.60% Croat
30.00% Bulgarians
03.80% Italy_Aosta_Valley
01.80% Italy_Trentino
00.60% Romanians
00.40% Swiss_Italian
00.40% Japanese
00.40% JPT30

Dodecad World9 (Admixture Studio)
54.00% Hungarians
36.40% Bulgarians_Y
06.20% N_Italian_D
02.00% Bulgarian_D
00.60% CDX30
00.40% French
00.40% Dai


G25 Results:

G25 Modern Averages Scaled
Target: Chris_scaled
Distance: 1.8234% / 0.01823434 | ADC: 0.25x
51.8 Moldovan
20.6 Bosnian
19.4 Montenegrin
3.2 Sardinian
2.2 Cossack_Ukrainian
1.0 Adygei
1.0 Li
0.8 Spanish_Soria

G25 Modern Averages Unscaled
Target: Chris
Distance: 1.0188% / 0.01018786 | ADC: 0.25x
25.8 Gagauz
18.8 Lithuanian_VA
14.6 Bosnian
12.8 Bulgarian
9.2 Montenegrin
7.6 Sardinian
5.0 Adygei
3.4 Cossack_Ukrainian
1.6 Ket
0.6 Spanish_La_Rioja
0.4 Ju_hoan_North
0.2 Li

Marques Gencove-like Calculator (very accurate imo):
Target: Chris_scaled
Distance: 1.9244% / 0.01924380 | ADC: 0.25x
65.8 East_Mediterranean:Bulgarian
28.0 East_European:Ukrainian
5.2 South_West_European:French_South
1.0 South_east_Asia:Dai


Someone help me lol I'm so confused xD

[17571imre]

me and my mother score it too. i thinks its just part of your Paleo_Balkan ancestry.

[Chris596]

me and my mother score it too. i thinks its just part of your Paleo_Balkan ancestry.

How much exactly? Is it this consistent like in my case? I've checked many Balkan samples (like Serbian, Macedonian, Bulgarian, Romanian, etc.) on different calculators (Eurogenes, MDLP, Dodecad, G25, etc.) and they don't score Italian/Sardinian, to be more precise, they score much much less than I score. This is why I want to do 23andme again :picard2: to see what they tell me.

Also, I don't think this is an error because it consistently pops up with the Hungarian/Balkan samples and it's not a little amount either.

[Ion Basescul]

I'm pretty sure you only have Hungarian admixture and the rest is just theories attempting to explain the said Southern-like genetic profile. Every population has a gradient, and not everyone will score like the average. You are Southern compared to average Hungarian and in Serbian range, Universe is Northern and in Ukrainian range. Both aren't any less Hungarian.

[Zoro]

Because 8000 year old Anatolian farmers were so successful in Europe and contributed so heavily to modern Europeans all Europeans carry their genes. It’s impossible for gedmatch calculators or G25 to filter out these ancient genes. They are simply not sophisticated enough.The other issue with those calculators and G25 is that they falsely make ethnic groups very homogeneous which they clearly are not. Formal stats have proved that individuals within ethnic groups are much more heterogeneous than these calculators and 23andme make them out to be

[Abriekman]

I'm pretty sure you only have Hungarian admixture and the rest is just theories attempting to explain the said Southern-like genetic profile. Every population has a gradient, and not everyone will score like the average. You are Southern compared to average Hungarian and in Serbian range, Universe is Northern and in Ukrainian range. Both aren't any less Hungarian.

Chris Hungarian ancestors did not mix with Germanic people. That is result of that

[vbnetkhio]

Because 8000 year old Anatolian farmers were so successful in Europe and contributed so heavily to modern Europeans all Europeans carry their genes. It’s impossible for gedmatch calculators or G25 to filter out these ancient genes.
you can come close if you use the right reference samples. eurogenes and g25 are based on modern populations, so they are not good for these very old ancestries.
MDLP calcs on the other hand actually have components like EEF and WHG.



They are simply not sophisticated enough.The other issue with those calculators and G25 is that they falsely make ethnic groups very homogeneous which they clearly are not. Formal stats have proved that individuals within ethnic groups are much more heterogeneous than these calculators and 23andme make them out to be

ethnic groups have huge ranges in gedmatch and g25 too, if you collect enough samples. for commercial companies, i agree. they do this on purpose because they are deisgned for mixed americans who want to know their old world ancestry.

[Chris596]

I'm pretty sure you only have Hungarian admixture and the rest is just theories attempting to explain the said Southern-like genetic profile. Every population has a gradient, and not everyone will score like the average. You are Southern compared to average Hungarian and in Serbian range, Universe is Northern and in Ukrainian range. Both aren't any less Hungarian.


Because 8000 year old Anatolian farmers were so successful in Europe and contributed so heavily to modern Europeans all Europeans carry their genes. It’s impossible for gedmatch calculators or G25 to filter out these ancient genes. They are simply not sophisticated enough.The other issue with those calculators and G25 is that they falsely make ethnic groups very homogeneous which they clearly are not. Formal stats have proved that individuals within ethnic groups are much more heterogeneous than these calculators and 23andme make them out to be


Chris Hungarian ancestors did not mix with Germanic people. That is result of that

Alright, these are all reasonable answers. I think that I might have some ,,Italian-like'' genetics, not directly from modern Italians. Something like a type of admixture, because of the close proximity of the Italian Peninsula and the Balkan Peninsula. This is just my theory, I could be totally wrong. Let's see what 23andme will say about this, I expect almost full Balkan results.

[Ion Basescul]

The other issue with those calculators and G25 is that they falsely make ethnic groups very homogeneous which they clearly are not.

Yep, Chris should ask Dunai, who has a lot of K13 individual Hungarian results, to run him against those. I'm sure that he'll get a very low distance.

[Ion Basescul]

Let's see what 23andme will say about this, I expect almost full Balkan results.

Since our results are similar, you should have expected about 50% South Euro, 45% East Euro and the rest in broadly with their previous version of the algorithm. I don't know how my results would look now as they haven't updated them, but I would guess about 70/30, and this is what you should also expect. The raw data however should be pretty similar to your MyHeritage's, so not much will change in practice.

[Zoro]

you can come close if you use the right reference samples. eurogenes and g25 are based on modern populations, so they are not good for these very old ancestries.
MDLP calcs on the other hand actually have components like EEF and WHG.



ethnic groups have huge ranges in gedmatch and g25 too, if you collect enough samples. for commercial companies, i agree. they do this on purpose because they are deisgned for mixed americans who want to know their old world ancestry.

That's not what I mean. What I meant is that in qpAdm or Dstats if you want to filter EEF genes you simply add EEF as an OUTGROUP. There's no place to add outgroups in admixture calculators or G25.
The other thing is qpAdm tells you if the model is a fail. Admixture or G25 doesn't so people assume all their models are good.

Also the individual variation within ethnic groups using admixture or G25 is not nearly as large as reality. You can do an experiment:

Do a Plink PCA based on 12 ancient samples and 1 test individual and see the coordinates for the individual. Next do run a PCA with the same ancients and 5 individuals from 1 ethnic group. You will see that 1 individual coordinates change. The other 4 from his ethnic group pull him towards them.

[Zoro]

you can come close if you use the right reference samples. eurogenes and g25 are based on modern populations, so they are not good for these very old ancestries.
MDLP calcs on the other hand actually have components like EEF and WHG.



ethnic groups have huge ranges in gedmatch and g25 too, if you collect enough samples. for commercial companies, i agree. they do this on purpose because they are deisgned for mixed americans who want to know their old world ancestry.

That's not what I mean. What I meant is that in qpAdm or Dstats if you want to filter EEF genes you simply add EEF as an OUTGROUP. There's no place to add outgroups in admixture calculators or G25.
The other thing is qpAdm tells you if the model is a fail. Admixture or G25 doesn't so people assume all their models are good.

Also the individual variation within ethnic groups using admixture or G25 is not nearly as large as reality. You can do an experiment:

Do a Plink PCA based on 12 ancient samples and 1 test individual and see the coordinates for the individual. Next do run a PCA with the same ancients and 5 individuals from 1 ethnic group. You will see that 1 individual coordinates change. The other 4 from his ethnic group pull him towards them.

[vbnetkhio]

That's not what I mean. What I meant is that in qpAdm or Dstats if you want to filter EEF genes you simply add EEF as an OUTGROUP. There's no place to add outgroups in admixture calculators or G25.
The other thing is qpAdm tells you if the model is a fail. Admixture or G25 doesn't so people assume all their models are good.

you are right, i thought by filter you meant tell you your exact amount of eef admixture.



Also the individual variation within ethnic groups using admixture or G25 is not nearly as large as reality. You can do an experiment:

Do a Plink PCA based on 12 ancient samples and 1 test individual and see the coordinates for the individual. Next do run a PCA with the same ancients and 5 individuals from 1 ethnic group. You will see that 1 individual coordinates change. The other 4 from his ethnic group pull him towards them.

if you project samples onto the pca instead of recalcualting the whole pca, this won't happen.
it's the same with admixture, you can first make a calculator in admixture, and then project samples into the model with DIYdodecad.

[CommonSense]

To me it seems like the result of the calculators just trying to give you the best possible fit (lowest distance). Of course, like with any possible admix the answers lie in your match list.

[Zoro]

you are right, i thought by filter you meant tell you your exact amount of eef admixture.



if you project samples onto the pca instead of recalcualting the whole pca, this won't happen.
it's the same with admixture, you can first make a calculator in admixture, and then project samples into the model with DIYdodecad.

Yes but G25 and most calculators were not made by using ancient samples only and only high quality ones. They were made using a mish mosh of ancient and modern samples. That's what I am referring to

[RyoHazuki]

Don't take them literally. It's G25 attempting to rationalize your southern shiftedness between using related populations and overfitting with any high EEF populations. Speaking personally, this is why I hand over fist get UK + Spain/Portugal over any French group in mixed models.

[Scandal]

Chris Hungarian ancestors did not mix with Germanic people. That is result of that

Yes. When we substract germanic admixture from Hungarians (around 25%) we get a 50% native balkanite 50% slavic population like Serbs. Hungarians can be modeled as 75% Serbs 25% Germans. As for me I have slovak ancestry which explain my dna test result.

[someonenotyou]

Alright, so I've noticed this a long time ago, but I can't explain it. I know that I'm already very Southern European (Southeastern), but I keep scoring smaller/bigger amounts of proper Southern European (like Sardinian/Italian/Spanish). Using my original Myheritage data and coordinates from GEDmatch, it appears on many different calculators. Also in Admixture Studio I score some Southern European.

If all of this is not enough, I also keep scoring some S. European on G25.

I have a confession to make: I know this is hilarious but when I posted my Admixture Studio results and some other results too, I always deleted the Southern European populations like a maniac, thinking that it can't be true so it won't appear in my results... So check this out:


What could be the reason behind it? Weirdly enough, I got Italian on FTDNA too


Vahaduo GEDmatch

Eurogenes EUtest
Target: Chris
Distance: 2.5116% / 2.51155634 | ADC: 0.25x
44.2 Serbian
35.4 Romanian
11.2 Hungarian
9.2 North_Italian

Eurogenes Jtest
Target: Chris
Distance: 2.3425% / 2.34247159 | ADC: 0.25x
47.8 HU
44.0 RO
5.2 North_Italian
2.4 AJ
0.6 Sardinian

HarappaWorld
Target: Chris
Distance: 1.2399% / 1.23985226 | ADC: 0.25x
80.2 Serbian
12.0 Italian
5.2 Romanian
2.0 Bulgarian
0.6 Japanese2

Dodecad K12b (Admixture Studio)
62.60% Croat
30.00% Bulgarians
03.80% Italy_Aosta_Valley
01.80% Italy_Trentino
00.60% Romanians
00.40% Swiss_Italian
00.40% Japanese
00.40% JPT30

Dodecad World9 (Admixture Studio)
54.00% Hungarians
36.40% Bulgarians_Y
06.20% N_Italian_D
02.00% Bulgarian_D
00.60% CDX30
00.40% French
00.40% Dai


G25 Results:

G25 Modern Averages Scaled
Target: Chris_scaled
Distance: 1.8234% / 0.01823434 | ADC: 0.25x
51.8 Moldovan
20.6 Bosnian
19.4 Montenegrin
3.2 Sardinian
2.2 Cossack_Ukrainian
1.0 Adygei
1.0 Li
0.8 Spanish_Soria

G25 Modern Averages Unscaled
Target: Chris
Distance: 1.0188% / 0.01018786 | ADC: 0.25x
25.8 Gagauz
18.8 Lithuanian_VA
14.6 Bosnian
12.8 Bulgarian
9.2 Montenegrin
7.6 Sardinian
5.0 Adygei
3.4 Cossack_Ukrainian
1.6 Ket
0.6 Spanish_La_Rioja
0.4 Ju_hoan_North
0.2 Li

Marques Gencove-like Calculator (very accurate imo):
Target: Chris_scaled
Distance: 1.9244% / 0.01924380 | ADC: 0.25x
65.8 East_Mediterranean:Bulgarian
28.0 East_European:Ukrainian
5.2 South_West_European:French_South
1.0 South_east_Asia:Dai


Someone help me lol I'm so confused xD

Since I'm also a southern shifting Hungarian. (Although not as much as you.) These are my results for comparisons sake.

EU Test
Target:
Distance: 4.7898% / 4.78984500 | ADC: 0.25x
40.6 SerbianBosnia
34.2 Norwegian
9.8 Moldavian_1
8.2 Sardinian
6.8 Lezgin
0.4 Kalash

Jtest
Target:
Distance: 5.6644% / 5.66440148 | ADC: 0.25x
37.4 Serbian
36.0 NO
9.8 Ukrainian-Russian
8.0 Sardinian
3.0 Lezgin
2.8 AJ
1.6 Komi
1.4 Kalash

Harappaworld
Target:
Distance: 1.4195% / 1.41946775 | ADC: 0.25x
48.2 Hungarian
45.8 French
2.0 Brahui
1.4 Makrani
1.4 Yaghnobi
1.2 Selkup

G25 Modern Scaled

Target:
Distance: 1.3983% / 0.01398279 | ADC: 0.25x
73.6 Austrian
14.8 Moldovan
6.4 Danish
4.2 Iranian_Mazandarani
0.8 Czech
0.2 Balochi


G25 Modern Unscaled

Target:
Distance: 0.6768% / 0.00676754 | ADC: 0.25x
48.2 Austrian
16.8 Danish
8.8 Bosnian
7.4 Czech
7.2 Slovenian
3.6 Sephardic_Jew
3.4 Yukagir_Forest
2.0 Iranian_Mazandarani
1.4 Even
1.0 Balochi
0.2 Roma_Porto

Marques Gencove-like
Target:
Distance: 1.6163% / 0.01616305 | ADC: 0.25x
59.6 North_and_central_European:German
24.4 East_Mediterranean:Bulgarian
11.8 East_European:Ukrainian
4.2 Anatolia_Caucasus_Iranian_Plateau:Iran

[JamesBond007]

Because 8000 year old Anatolian farmers were so successful in Europe and contributed so heavily to modern Europeans all Europeans carry their genes. It’s impossible for gedmatch calculators or G25 to filter out these ancient genes. They are simply not sophisticated enough.The other issue with those calculators and G25 is that they falsely make ethnic groups very homogeneous which they clearly are not. Formal stats have proved that individuals within ethnic groups are much more heterogeneous than these calculators and 23andme make them out to be

G25 is too fine-grained for modern European populations and that is not its only problem. I know for fact I'm Scottish-American but check out what G25 modern tells me :

Distance to: Kevin_scaled
0.02865262 Dutch
0.02966312 Norwegian
0.03097093 Welsh
0.03128912 Irish
0.03153596 Scottish
0.03173767 English
0.03264251 Orcadian

Distance: 2.4032% / 0.02403165 | ADC: 0.25x
72.0 Norwegian
14.0 Spanish_Soria
11.10 Dutch
2.0 Tajik_Shugnan

I have no Spanish blood it is picking up the ancient neolithic-mesolithic megalithic stone henge culture from Britain there and translating into 'Spanish' or something.

[Zoro]

G25 is too fine-grained for modern European populations and that is not its only problem. I know for fact I'm Scottish-American but check out what G25 modern tells me :

Distance to: Kevin_scaled
0.02865262 Dutch
0.02966312 Norwegian
0.03097093 Welsh
0.03128912 Irish
0.03153596 Scottish
0.03173767 English
0.03264251 Orcadian

Distance: 2.4032% / 0.02403165 | ADC: 0.25x
72.0 Norwegian
14.0 Spanish_Soria
11.10 Dutch
2.0 Tajik_Shugnan

I have no Spanish blood it is picking up the ancient neolithic-mesolithic megalithic stone henge culture from Britain there and translating into 'Spanish' or something.

Good example. The problem is also that people on forums have been so brainwashed by the results of these types of calculators that it has become reality to them so now if they see their real results from more scientifically acceptable methods they think something is wrong. With moderns you may get an idea something maybe off but with ancient modeling people would be clueless and they simply accept the results. Basically a disservice

[RyoHazuki]

G25 is too fine-grained for modern European populations and that is not its only problem. I know for fact I'm Scottish-American but check out what G25 modern tells me :

Distance to: Kevin_scaled
0.02865262 Dutch
0.02966312 Norwegian
0.03097093 Welsh
0.03128912 Irish
0.03153596 Scottish
0.03173767 English
0.03264251 Orcadian

Distance: 2.4032% / 0.02403165 | ADC: 0.25x
72.0 Norwegian
14.0 Spanish_Soria
11.10 Dutch
2.0 Tajik_Shugnan

I have no Spanish blood it is picking up the ancient neolithic-mesolithic megalithic stone henge culture from Britain there and translating into 'Spanish' or something.

It only shows Spain because the main component is a very northern shifted one, just beyond the northernmost Scottish people. Model yourself any other way and it disappears.

[Scandal]

I'm pretty sure you only have Hungarian admixture and the rest is just theories attempting to explain the said Southern-like genetic profile. Every population has a gradient, and not everyone will score like the average. You are Southern compared to average Hungarian and in Serbian range, Universe is Northern and in Ukrainian range. Both aren't any less Hungarian.

Well, he's also half Székely. So vlach/romanian admixture is a possibility too.

[vbnetkhio]

Yes but G25 and most calculators were not made by using ancient samples only and only high quality ones. They were made using a mish mosh of ancient and modern samples. That's what I am referring to

i think g25 was made just from modern samples, and ancients were projected into it.