Balto-Slavic scale

**cass** · 05-27-2020, 12:38 PM

Target: Sorb_Niederlausitz
Distance: 1.9298% / 0.01929826 | ADC: 0.25x

67.0 Slavic
18.2 Baltic
14.8 Balkan

0 Germanic?
It seems to be wrongly calibrated.

**Jana** · 05-27-2020, 12:44 PM

Originally Posted by cass

Target: Sorb_Niederlausitz
Distance: 1.9298% / 0.01929826 | ADC: 0.25x

67.0 Slavic
18.2 Baltic
14.8 Balkan

0 Germanic?
It seems to be wrongly calibrated.

They score no similarity with medieval Germanic samples, sorry. Not weird considering they plot east of Czechs.
However Balkan samples have some western european drift, especially those from Croatia.

My guess is David handpicked purest Sorbs for his average he could find. His Polish average is also very Slavic and eastern shifted.

**~~Skiz-gaaR~~** · 07-17-2020, 04:50 AM

Originally Posted by Feiichy

WEST SLAVS

Target: Polish
Distance: 1.0294% / 0.01029432 | ADC: 0.25x

86.0 Slavic
9.6 Germanic
4.4 Baltic

Target: Slovakian
Distance: 1.9145% / 0.01914454 | ADC: 0.25x

84.2 Slavic
10.8 Balkan
4.2 Germanic
0.8 Baltic

Target: Sorb_Niederlausitz
Distance: 1.9298% / 0.01929826 | ADC: 0.25x

67.0 Slavic
18.2 Baltic
14.8 Balkan

Target: Czech
Distance: 1.1885% / 0.01188460 | ADC: 0.25x

56.6 Slavic
34.2 Germanic
5.2 Balkan
2.6 Greco-Roman
1.4 Celtic

How do Slovaks and Sorbs have Balkan but Poles don't? Sorbs are further west than Poles

**-Scar-** · 08-15-2020, 11:03 PM

Originally Posted by Stearsolina

WEST SLAVS

Target: Polish
Distance: 1.0294% / 0.01029432 | ADC: 0.25x

86.0 Slavic
9.6 Germanic
4.4 Baltic

Best model based on historical data.
Before the Slavic migration the people in Western Poland were Swedish-like, clearly there was a large scale replacement.

**Token** · 08-17-2020, 01:18 PM

Originally Posted by Stearsolina

Target: Hungarian
Distance: 0.7988% / 0.00798832 | ADC: 0.25x

52.8 Slavic
23.8 Germanic
21.8 Balkan
1.6 Greco-Roman

I've been playing with this Hungarian academic average with qpAdm. Apparently they don't need anything besides Germanic and Slavic, the Balkan stuff is just Global25 overfitting the data. This is one of the models that i ran (obs: qpAdm does not distinguish North Europeans, so Germany_EMedieval is a stand in for both Slavic and Germanic. The exact proportions of both should be analysed using G25):

Hungarian
Germany_EMedieval 0.873 +- 0.100
Lech_MBA 0.127 +- 0.100
chisq 10.154
tail prob 0 516617

So Lech_MBA isn't required (and neither are Hungary_Scythian and Croatia_EIA) since the std.error is basically equal to the admixture coefficient, so the nested model with only Germany_EMedieval (p-value: 0.14) is closer to optimal.

Full output:

Code:

left pops:
Hungarian
Germany_EMedieval.SG
Germany_Lech_MBA

right pops:
Czech_Vestonice16
Russia_Ust_Ishim.DG
Russia_MA1_HG.SG
Italy_North_Villabruna_HG
Jordan_PPNB
Romania_Mesolithic_IronGates
Georgia_Satsurblia.SG
Morocco_Iberomaurusian
Iran_GanjDareh_N
Cameroon_SMA.DG
Anatolia_N
Russia_Steppe_Eneolithic
Russia_HG_Karelia

  0            Hungarian   20
  1 Germany_EMedieval.SG   35
  2     Germany_Lech_MBA    7
  3    Czech_Vestonice16    1
  4  Russia_Ust_Ishim.DG    1
  5     Russia_MA1_HG.SG    1
  6 Italy_North_Villabruna_HG    1
  7          Jordan_PPNB    4
  8 Romania_Mesolithic_IronGates    4
  9 Georgia_Satsurblia.SG    1
 10 Morocco_Iberomaurusian    6
 11     Iran_GanjDareh_N    8
 12      Cameroon_SMA.DG    1
 13           Anatolia_N   22
 14 Russia_Steppe_Eneolithic    3
 15    Russia_HG_Karelia    2
jackknife block size:     0.050
snps: 593124  indivs: 117
number of blocks for block jackknife: 711
dof (jackknife):   602.170
numsnps used: 46474
codimension 1
f4info: 
f4rank: 1 dof:     11 chisq:    10.154 tail:           0.51661742 dofdiff:     13 chisqdiff:   -10.154 taildiff:                    1
B:
          scale     1.000 
Russia_Ust_Ishim.DG     1.491 
Russia_MA1_HG.SG     1.268 
Italy_North_Villabruna_HG     0.060 
    Jordan_PPNB    -0.175 
Romania_Mesolithic_IronGates     1.095 
Georgia_Satsurblia.SG     0.757 
Morocco_Iberomaurusian     0.364 
Iran_GanjDareh_N     0.759 
Cameroon_SMA.DG     1.266 
     Anatolia_N     0.177 
Russia_Steppe_Eneolithic     1.180 
Russia_HG_Karelia     1.621 
A:
          scale   941.312 
Germany_EMedieval.SG     0.204 
Germany_Lech_MBA    -1.399 


full rank 1
f4info: 
f4rank: 2 dof:      0 chisq:     0.000 tail:                    1 dofdiff:     11 chisqdiff:    10.154 taildiff:           0.51661742
B:
          scale     1.000     1.000 
Russia_Ust_Ishim.DG     1.520    -0.475 
Russia_MA1_HG.SG     1.145     0.221 
Italy_North_Villabruna_HG     0.102     0.032 
    Jordan_PPNB    -0.226     1.608 
Romania_Mesolithic_IronGates     1.080     0.028 
Georgia_Satsurblia.SG     0.674     1.886 
Morocco_Iberomaurusian     0.357     1.491 
Iran_GanjDareh_N     0.776     0.505 
Cameroon_SMA.DG     1.309    -0.603 
     Anatolia_N     0.145     1.543 
Russia_Steppe_Eneolithic     1.170     0.137 
Russia_HG_Karelia     1.692    -0.584 
A:
          scale   991.102  3066.804 
Germany_EMedieval.SG     0.674     1.243 
Germany_Lech_MBA    -1.243     0.674 


best coefficients:     0.873     0.127 
ssres:
    0.000479023     0.000581661     0.000056186     0.000515624     0.000479649     0.001012712     0.000723984     0.000529943     0.000338111     0.000651928     0.000560565     0.000512255 
    0.828304055     1.005779384     0.097153245     0.891591578     0.829385221     1.751132727     1.251877553     0.916352117     0.584644749     1.127281671     0.969301143     0.885767176 

Jackknife mean:      0.864886518     0.135113482 
      std. errors:     0.100     0.100 

error covariance (* 1000000)
     10007     -10007 
    -10007      10007 


    fixed pat  wt  dof     chisq       tail prob
           00  0    11    10.154        0.516617     0.873     0.127 
           01  1    12    12.339        0.418882     1.000    -0.000 
           10  1    12    41.981     3.35445e-05     0.000     1.000 
best pat:           00         0.516617              -  -
best pat:           01         0.418882  chi(nested):     2.185 p-value for nested model:        0.139368

**Aren** · 08-17-2020, 01:33 PM

Originally Posted by Token

I've been playing with this Hungarian academic average with qpAdm. Apparently they don't need anything besides Germanic and Slavic, the Balkan stuff is just Global25 overfitting the data. This is one of the models that i ran (obs: qpAdm does not distinguish North Europeans, so Germany_EMedieval is a stand in for both Slavic and Germanic. The exact proportions of both should be analysed using G25):

Hungarian
Germany_EMedieval 0.873 +- 0.100
Lech_MBA 0.127 +- 0.100
chisq 10.154
tail prob 0 516617

So Lech_MBA isn't required (and neither are Hungary_Scythian and Croatia_EIA) since the std.error is basically equal to the admixture coefficient, so the nested model with only Germany_EMedieval (p-value: 0.14) is closer to optimal.

Full output:

Code:

left pops:
Hungarian
Germany_EMedieval.SG
Germany_Lech_MBA

right pops:
Czech_Vestonice16
Russia_Ust_Ishim.DG
Russia_MA1_HG.SG
Italy_North_Villabruna_HG
Jordan_PPNB
Romania_Mesolithic_IronGates
Georgia_Satsurblia.SG
Morocco_Iberomaurusian
Iran_GanjDareh_N
Cameroon_SMA.DG
Anatolia_N
Russia_Steppe_Eneolithic
Russia_HG_Karelia

  0            Hungarian   20
  1 Germany_EMedieval.SG   35
  2     Germany_Lech_MBA    7
  3    Czech_Vestonice16    1
  4  Russia_Ust_Ishim.DG    1
  5     Russia_MA1_HG.SG    1
  6 Italy_North_Villabruna_HG    1
  7          Jordan_PPNB    4
  8 Romania_Mesolithic_IronGates    4
  9 Georgia_Satsurblia.SG    1
 10 Morocco_Iberomaurusian    6
 11     Iran_GanjDareh_N    8
 12      Cameroon_SMA.DG    1
 13           Anatolia_N   22
 14 Russia_Steppe_Eneolithic    3
 15    Russia_HG_Karelia    2
jackknife block size:     0.050
snps: 593124  indivs: 117
number of blocks for block jackknife: 711
dof (jackknife):   602.170
numsnps used: 46474
codimension 1
f4info: 
f4rank: 1 dof:     11 chisq:    10.154 tail:           0.51661742 dofdiff:     13 chisqdiff:   -10.154 taildiff:                    1
B:
          scale     1.000 
Russia_Ust_Ishim.DG     1.491 
Russia_MA1_HG.SG     1.268 
Italy_North_Villabruna_HG     0.060 
    Jordan_PPNB    -0.175 
Romania_Mesolithic_IronGates     1.095 
Georgia_Satsurblia.SG     0.757 
Morocco_Iberomaurusian     0.364 
Iran_GanjDareh_N     0.759 
Cameroon_SMA.DG     1.266 
     Anatolia_N     0.177 
Russia_Steppe_Eneolithic     1.180 
Russia_HG_Karelia     1.621 
A:
          scale   941.312 
Germany_EMedieval.SG     0.204 
Germany_Lech_MBA    -1.399 


full rank 1
f4info: 
f4rank: 2 dof:      0 chisq:     0.000 tail:                    1 dofdiff:     11 chisqdiff:    10.154 taildiff:           0.51661742
B:
          scale     1.000     1.000 
Russia_Ust_Ishim.DG     1.520    -0.475 
Russia_MA1_HG.SG     1.145     0.221 
Italy_North_Villabruna_HG     0.102     0.032 
    Jordan_PPNB    -0.226     1.608 
Romania_Mesolithic_IronGates     1.080     0.028 
Georgia_Satsurblia.SG     0.674     1.886 
Morocco_Iberomaurusian     0.357     1.491 
Iran_GanjDareh_N     0.776     0.505 
Cameroon_SMA.DG     1.309    -0.603 
     Anatolia_N     0.145     1.543 
Russia_Steppe_Eneolithic     1.170     0.137 
Russia_HG_Karelia     1.692    -0.584 
A:
          scale   991.102  3066.804 
Germany_EMedieval.SG     0.674     1.243 
Germany_Lech_MBA    -1.243     0.674 


best coefficients:     0.873     0.127 
ssres:
    0.000479023     0.000581661     0.000056186     0.000515624     0.000479649     0.001012712     0.000723984     0.000529943     0.000338111     0.000651928     0.000560565     0.000512255 
    0.828304055     1.005779384     0.097153245     0.891591578     0.829385221     1.751132727     1.251877553     0.916352117     0.584644749     1.127281671     0.969301143     0.885767176 

Jackknife mean:      0.864886518     0.135113482 
      std. errors:     0.100     0.100 

error covariance (* 1000000)
     10007     -10007 
    -10007      10007 


    fixed pat  wt  dof     chisq       tail prob
           00  0    11    10.154        0.516617     0.873     0.127 
           01  1    12    12.339        0.418882     1.000    -0.000 
           10  1    12    41.981     3.35445e-05     0.000     1.000 
best pat:           00         0.516617              -  -
best pat:           01         0.418882  chi(nested):     2.185 p-value for nested model:        0.139368

Your right pops are too old and basic for such a recent run(migration era samples).
There’s absolutely no way Hungarians don’t have any more Southern ancestry than the Belarussian-like early Slavs and the Scandinavian-like early Germanics, Hungarians score significantly more Barcin than both these populations. There’s some East Eurasian there too albeit noise level.

**knez01** · 08-17-2020, 01:50 PM

Target: Knez_scaled
Distance: 2.2982% / 0.02298201
33.2 Slavic
31.8 Greco-Roman
23.0 Baltic
12.0 Balkan

Target: Knez_scaled
Distance: 2.4368% / 0.02436807 | ADC: 0.25x
66.4 Slavic
23.0 Greco-Roman
10.6 Balkan

**vbnetkhio** · 08-17-2020, 01:51 PM

Originally Posted by Token

I've been playing with this Hungarian academic average with qpAdm. Apparently they don't need anything besides Germanic and Slavic, the Balkan stuff is just Global25 overfitting the data. This is one of the models that i ran (obs: qpAdm does not distinguish North Europeans, so Germany_EMedieval is a stand in for both Slavic and Germanic. The exact proportions of both should be analysed using G25):

Hungarian
Germany_EMedieval 0.873 +- 0.100
Lech_MBA 0.127 +- 0.100
chisq 10.154
tail prob 0 516617

So Lech_MBA isn't required (and neither are Hungary_Scythian and Croatia_EIA) since the std.error is basically equal to the admixture coefficient, so the nested model with only Germany_EMedieval (p-value: 0.14) is closer to optimal.

Full output:

Code:

left pops:
Hungarian
Germany_EMedieval.SG
Germany_Lech_MBA

right pops:
Czech_Vestonice16
Russia_Ust_Ishim.DG
Russia_MA1_HG.SG
Italy_North_Villabruna_HG
Jordan_PPNB
Romania_Mesolithic_IronGates
Georgia_Satsurblia.SG
Morocco_Iberomaurusian
Iran_GanjDareh_N
Cameroon_SMA.DG
Anatolia_N
Russia_Steppe_Eneolithic
Russia_HG_Karelia

  0            Hungarian   20
  1 Germany_EMedieval.SG   35
  2     Germany_Lech_MBA    7
  3    Czech_Vestonice16    1
  4  Russia_Ust_Ishim.DG    1
  5     Russia_MA1_HG.SG    1
  6 Italy_North_Villabruna_HG    1
  7          Jordan_PPNB    4
  8 Romania_Mesolithic_IronGates    4
  9 Georgia_Satsurblia.SG    1
 10 Morocco_Iberomaurusian    6
 11     Iran_GanjDareh_N    8
 12      Cameroon_SMA.DG    1
 13           Anatolia_N   22
 14 Russia_Steppe_Eneolithic    3
 15    Russia_HG_Karelia    2
jackknife block size:     0.050
snps: 593124  indivs: 117
number of blocks for block jackknife: 711
dof (jackknife):   602.170
numsnps used: 46474
codimension 1
f4info: 
f4rank: 1 dof:     11 chisq:    10.154 tail:           0.51661742 dofdiff:     13 chisqdiff:   -10.154 taildiff:                    1
B:
          scale     1.000 
Russia_Ust_Ishim.DG     1.491 
Russia_MA1_HG.SG     1.268 
Italy_North_Villabruna_HG     0.060 
    Jordan_PPNB    -0.175 
Romania_Mesolithic_IronGates     1.095 
Georgia_Satsurblia.SG     0.757 
Morocco_Iberomaurusian     0.364 
Iran_GanjDareh_N     0.759 
Cameroon_SMA.DG     1.266 
     Anatolia_N     0.177 
Russia_Steppe_Eneolithic     1.180 
Russia_HG_Karelia     1.621 
A:
          scale   941.312 
Germany_EMedieval.SG     0.204 
Germany_Lech_MBA    -1.399 


full rank 1
f4info: 
f4rank: 2 dof:      0 chisq:     0.000 tail:                    1 dofdiff:     11 chisqdiff:    10.154 taildiff:           0.51661742
B:
          scale     1.000     1.000 
Russia_Ust_Ishim.DG     1.520    -0.475 
Russia_MA1_HG.SG     1.145     0.221 
Italy_North_Villabruna_HG     0.102     0.032 
    Jordan_PPNB    -0.226     1.608 
Romania_Mesolithic_IronGates     1.080     0.028 
Georgia_Satsurblia.SG     0.674     1.886 
Morocco_Iberomaurusian     0.357     1.491 
Iran_GanjDareh_N     0.776     0.505 
Cameroon_SMA.DG     1.309    -0.603 
     Anatolia_N     0.145     1.543 
Russia_Steppe_Eneolithic     1.170     0.137 
Russia_HG_Karelia     1.692    -0.584 
A:
          scale   991.102  3066.804 
Germany_EMedieval.SG     0.674     1.243 
Germany_Lech_MBA    -1.243     0.674 


best coefficients:     0.873     0.127 
ssres:
    0.000479023     0.000581661     0.000056186     0.000515624     0.000479649     0.001012712     0.000723984     0.000529943     0.000338111     0.000651928     0.000560565     0.000512255 
    0.828304055     1.005779384     0.097153245     0.891591578     0.829385221     1.751132727     1.251877553     0.916352117     0.584644749     1.127281671     0.969301143     0.885767176 

Jackknife mean:      0.864886518     0.135113482 
      std. errors:     0.100     0.100 

error covariance (* 1000000)
     10007     -10007 
    -10007      10007 


    fixed pat  wt  dof     chisq       tail prob
           00  0    11    10.154        0.516617     0.873     0.127 
           01  1    12    12.339        0.418882     1.000    -0.000 
           10  1    12    41.981     3.35445e-05     0.000     1.000 
best pat:           00         0.516617              -  -
best pat:           01         0.418882  chi(nested):     2.185 p-value for nested model:        0.139368

Hungarians are a Nordo-Slavic mix like northeast Germans? this doesn't make sense.

could you try it with these outgroups?

set 1: Anatolia_N (25), CHG (2), EHG (4), ElMiron (1), Iran_Ganj_Dareh_N (3),
Jordan_PPNB (1), MA1 (1), Mbuti (10), Natufian (6), Ust_Ishim (1), Vestonice16 (1), WHG (6),
Russia_Yamnaya_Samara (9)

set 2: Anatolia_N (25), CHG (2), EHG (4), ElMiron (1), GoyetQ116-1 (1), Iran_Ganj_Dareh_N (3),
Jordan_PPNB (1), Kostenki14 (1), MA1 (1), Morocco_Iberomaurusian (6), Mota (1), Natufian (6),
Ust_Ishim (1), Vestonice16 (1), Italy_Villabruna (1), WHG (6), Russia_Yamnaya_Samara (9).

that's what was used in the Rome study

**Chris596** · 08-17-2020, 01:54 PM

Target: Chris_scaled
Distance: 1.9340% / 0.01933960 | ADC: 0.25x
45.8 Slavic
41.0 Balkan
9.2 Celtic
4.0 Greco-Roman

Target: ChrisFather
Distance: 4.3609% / 0.04360857 | ADC: 0.25x
50.6 Balkan
33.6 Slavic
15.8 Greco-Roman

**Token** · 08-17-2020, 01:59 PM

Originally Posted by Aren

Your right pops are too old and basic for such a recent run(migration era samples).
There’s absolutely no way Hungarians don’t have any more Southern ancestry than the Belarussian-like early Slavs and the Scandinavian-like early Germanics, Hungarians score significantly more Barcin than both these populations. There’s some East Eurasian there too albeit noise level.

No, my right pops are very solid for my purpose, have you bothered reading Patterson's notes or looking at the supp. files of genetic papers? Actually i am being quite restrictive here.

Your observations are all based on Global25 so they are meaningless for formal stats, run your qpAdm model and prove me wrong (spoiler: they will show the same). There is no East Eurasian in this Hungarian sample, albeit i am pretty sure Hungarians from the Great Plain will show some.

Thread: Balto-Slavic scale

Thread Tools

Thread Information

Users Browsing this Thread

Bookmarks

Bookmarks

Posting Permissions