Search This Blog

Sunday, 10 December 2023

Attempting to improve the qpAdm models from the new study: A genetic history of the Balkans from Roman frontier to Slavic migrations

Related post: Modelling the Roman era sample from Marathon Greece on qpAdm and G25

Study: A genetic history of the Balkans from Roman frontier to Slavic migrations

Post by Davidski about the study: https://eurogenes.blogspot.com/2024/01/romans-and-slavs-in-balkans-olalde-et.html

I used the reference/right populations as in that study but expanded upon them to decrease standard errors. If you have any criticism then tell me in the comments so i can improve the models.

The Balkan study didn't use a Balkan proxy for mainland Greeks even though they have a lot of Balkan Y-DNA from both sides of the Balkans, Arvanites/Albanians, Thracian etc, obviously in my models i have to take into account the Balkan admix. The models with South Slavic fit the best when rotating because they take into account the Balkan admx. They could have used a Balkan or South Slavic proxy if they added more right/reference populations + a lot of good quality Rome Imperial samples that are identical to Mugla West Anatolian samples to decrease the Standard Errors.

The "Ottoman Turkish" proxy of the study is also sus, the paper claims that it is central Asian mixed but i can't find any models showcasing how much Central Asian admix it has. My qpAdm models don't fail without Central Asian/East Asian despite using Mongolia_North_N in the reference/right pops. It seems that they might have used some very low Turkic mixed sample that is mostly Byzantine Greek and just tagged it as Ottoman Turkish. If i somehow missed their models about the Turkic admix of the "Ottoman Turkish" samples then tell me in the comments.

If you are curious about the absence of Mycenaean as a source in the qpAdm chart, please scroll down to the qpAdm reference section below for a detailed overview of the models utilized. The selection criteria prioritized models based on their P-values, with preference given to those demonstrating the highest statistical significance. Furthermore, modern Greeks scored less than 5% excess Mycenaean when using the Roman era Aegean proxy.

The Aegean proxy in my analysis consist of samples from the Roman-era Mugla in West Anatolia, along with Imperial Rome samples that are identical to those from Mugla. Furthermore, the Marathon sample from 300 AD southern Greece is identical to those samples. It scores a total of 30-50% Mycenaean admix depending on the proxies used so Mugla samples might be an actual good representative of Roman era Southern Greeks until we get more samples.

Explaining more about the Roman era Aegean proxy here:

More work will be be done on these models, i will update them in the future.

qpAdm models:

Scroll down to see all the models i ran.

Low P values (<0.01 / lower than 0.01) indicate poor fit of the tested model
Keep in mind that because you cant use too many overlapping sources on qpAdm then some sources are forced to overcompensate for the lacking sources.

Sample of the Roman era Greek samples might be faulty, from one model i've made without them, the result was a bit closer to the G25 charts. I will remake the chart another time.

I rotated the sources
qpAdm references/outgroups:
right = c('Mbuti.HO', 'Russia_Samara_EBA_Yamnaya', 'Russia_Karelia_HG', 'Serbia_IronGates_Mesolithic', 'Turkey_N', 'Iran_GanjDareh_N', 'Spain_IA', 'Greece_Minoan_Lassithi', 'Croatia_MLBA', 'Netherlands_EIA', 'Netherlands_MBA', 'Russia_IA_Ingria.SG', 'Latvia_BA', 'Turkey_EBA_II.SG', 'Israel_C', 'Armenia_EBA_KuraAraxes', 'Mongolia_North_N', 'Russia_MLBA_Sintashta', 'Iran_C_SehGabi', 'Tajikistan_Ksirov_Kushan', 'Croatia_EIA', 'Albania_BA_IA','Macedonia_Classical_Hellenistic', 'Bulgaria_EIA', 'Greece_Delphi_BA_Mycenaean', 'Greece_BA_Mycenaean', 'Lebanon_MBA.SG', 'Russia_Sunghir_Medieval.SG', 'Montenegro_Doclea_Roman.SG')

Aegean 1-200AD samples used https://pastebin.com/KdHE0uwT

G25 models:

I think G25 might be a bit better at making more complicated models and capturing small recent variations so check this post:
The qpAdm models above though confirm the main shifts that we see on G25


Source populations:


The Tajik like ancestry appear in the Levant during the Hellenistic and it also shows up in Cypriots

3 out of 8 of the Cypriot samples are outliers, 2 seem to be Maronite Cypriots while 1 is half mainland Greek. I don't know why they included them. I have seen dozens of Cypriot Greek samples from people whose all grandparents are native Cypriots and none of them were like those 3 samples. The Cypriot_Maronite_o samples can be modelled as 100% Lebanese Christian while the normal Cypriot Greek samples cannot.

P 0.0448 good fit
 target              left                  weight       se       z
  <chr>               <chr>                  <dbl>    <dbl>   <dbl>
1 Cypriot_Maronite_o.HO Lebanese_Christian.HO      1 1.14e-13 8.79e12

P 0.000000388 rejected
  target     left                  weight       se       z
  <chr>      <chr>                  <dbl>    <dbl>   <dbl>
1 Cypriot.HO Lebanese_Christian.HO      1 1.14e-13 8.79e12

P 0.887 good fit
  target                   left                  weight     se     z
  <chr>                    <chr>                  <dbl>  <dbl> <dbl>
1 Cypriot_mainlandmixed_o.HO Cypriot.HO             0.682 0.0877  7.78
2 Cypriot_mainlandmixed_o.HO Greek_Thessaloniki.HO  0.318 0.0877  3.62

P 0.919 good fit
  target                   left                      weight    se     z
  <chr>                    <chr>                      <dbl> <dbl> <dbl>
1 Cypriot_mainlandmixed_o.HO Cypriot.HO                 0.602 0.119  5.04
2 Cypriot_mainlandmixed_o.HO Greek_Athens_10-20.WGA.HO  0.398 0.119  3.33

Cypriot outlier samples:
Cyprus2AJ19.HO M Cypriot_Maronite_o.HO
Cyprus21AJ19.HO M Cypriot_Maronite_o.HO
Cyprus13AJ19.HO M Cypriot_mainlandmixed_o.HO

qpAdm References/outgroups:

Important info:
I rotated the sources/references unless i specified otherwise
---------------------------------------------------------------
right = c('Mbuti.HO', 'Russia_Samara_EBA_Yamnaya', 'Russia_Karelia_HG', 'Serbia_IronGates_Mesolithic', 'Turkey_N', 'Iran_GanjDareh_N', 'Spain_IA', 'Greece_Minoan_Lassithi', 'Croatia_MLBA', 'Netherlands_EIA', 'Netherlands_MBA', 'Russia_IA_Ingria.SG', 'Latvia_BA', 'Turkey_EBA_II.SG', 'Israel_C', 'Armenia_EBA_KuraAraxes', 'Mongolia_North_N', 'Russia_MLBA_Sintashta', 'Iran_C_SehGabi', 'Tajikistan_Ksirov_Kushan', 'Croatia_EIA', 'Albania_BA_IA','Macedonia_Classical_Hellenistic', 'Bulgaria_EIA', 'Greece_Delphi_BA_Mycenaean', 'Greece_BA_Mycenaean', 'Lebanon_MBA.SG', 'Russia_Sunghir_Medieval.SG', 'Montenegro_Doclea_Roman.SG')

left = c('Aegean(Roman_era_1-200AD).SG', 'Greece_BA_Mycenaean', 'Montenegro_Doclea_Roman.SG')

target = c('Greek_Athens_10-20.WGA.HO')

results = qpadm(prefix, left, right, target, allsnps = TRUE)
results$weights
results$popdrop


P 0.204
  target                left                            weight     se      z
  <chr>                 <chr>                            <dbl>  <dbl>  <dbl>
1 Greek_Thessaloniki.HO Aegean(Roman_era_1-200AD).SG    0.441  0.0378 11.7  
2 Greek_Thessaloniki.HO Macedonia_Classical_Hellenistic 0.0355 0.0560  0.633
3 Greek_Thessaloniki.HO Montenegro_Doclea_Roman.SG      0.523  0.0303 17.3 

P 0.136
  target                left                         weight     se     z
  <chr>                 <chr>                         <dbl>  <dbl> <dbl>
1 Greek_Thessaloniki.HO Aegean(Roman_era_1-200AD).SG  0.462 0.0196  23.6
2 Greek_Thessaloniki.HO Montenegro_Doclea_Roman.SG    0.538 0.0196  27.5

P 0.0771
  target                left                          weight     se      z
  <chr>                 <chr>                          <dbl>  <dbl>  <dbl>
1 Greek_Thessaloniki.HO Aegean(Roman_era_1-200AD).SG  0.478  0.0441 10.8  
2 Greek_Thessaloniki.HO Greece_BA_Mycenaean          -0.0186 0.0434 -0.429
3 Greek_Thessaloniki.HO Montenegro_Doclea_Roman.SG    0.541  0.0202 26.8

P 0.0759
  target                left                          weight     se      z
  <chr>                 <chr>                          <dbl>  <dbl>  <dbl>
1 Greek_Thessaloniki.HO Aegean(Roman_era_1-200AD).SG  0.478  0.0322 14.8  
2 Greek_Thessaloniki.HO Albania_BA_IA                -0.0343 0.0535 -0.642
3 Greek_Thessaloniki.HO Montenegro_Doclea_Roman.SG    0.557  0.0347 16.1  

P 0.0687
  target                left                         weight     se      z
  <chr>                 <chr>                         <dbl>  <dbl>  <dbl>
1 Greek_Thessaloniki.HO Aegean(Roman_era_1-200AD).SG 0.444  0.0335 13.3  
2 Greek_Thessaloniki.HO Bulgaria_EIA                 0.0235 0.0336  0.697
3 Greek_Thessaloniki.HO Montenegro_Doclea_Roman.SG   0.533  0.0203 26.3

P 0.0850
  <chr>                 <chr>                         <dbl>  <dbl>  <dbl>
1 Greek_Thessaloniki.HO Aegean(Roman_era_1-200AD).SG 0.578  0.0499 11.6  
2 Greek_Thessaloniki.HO Greece_BA_Mycenaean          0.0140 0.0488  0.286
3 Greek_Thessaloniki.HO Russia_Sunghir_Medieval.SG   0.408  0.0183 22.3

P 0.112
  target                left                            weight     se      z
  <chr>                 <chr>                            <dbl>  <dbl>  <dbl>
1 Greek_Thessaloniki.HO Aegean(Roman_era_1-200AD).SG    0.555  0.0447 12.4  
2 Greek_Thessaloniki.HO Macedonia_Classical_Hellenistic 0.0545 0.0611  0.892
3 Greek_Thessaloniki.HO Russia_Sunghir_Medieval.SG      0.390  0.0267 14.6
---------------------------------------------------------------------------------------

P 0.150
  target                    left                         weight     se     z
  <chr>                     <chr>                         <dbl>  <dbl> <dbl>
1 Greek_Athens_10-20.WGA.HO Aegean(Roman_era_1-200AD).SG  0.590 0.0216  27.3
2 Greek_Athens_10-20.WGA.HO Montenegro_Doclea_Roman.SG    0.410 0.0216  19.0

P 0.118
  target                    left                         weight     se      z
  <chr>                     <chr>                         <dbl>  <dbl>  <dbl>
1 Greek_Athens_10-20.WGA.HO Aegean(Roman_era_1-200AD).SG 0.676  0.0420 16.1  
2 Greek_Athens_10-20.WGA.HO Albania_BA_IA                0.0334 0.0662  0.504
3 Greek_Athens_10-20.WGA.HO Russia_Sunghir_Medieval.SG   0.291  0.0337  8.64 

P 0.0844
  target                    left                            weight     se      z
  <chr>                     <chr>                            <dbl>  <dbl>  <dbl>
1 Greek_Athens_10-20.WGA.HO Aegean(Roman_era_1-200AD).SG    0.652  0.0500 13.0  
2 Greek_Athens_10-20.WGA.HO Macedonia_Classical_Hellenistic 0.0620 0.0715  0.868
3 Greek_Athens_10-20.WGA.HO Russia_Sunghir_Medieval.SG      0.286  0.0308  9.30

P 0.0881
  target                    left                            weight     se      z
  <chr>                     <chr>                            <dbl>  <dbl>  <dbl>
1 Greek_Athens_10-20.WGA.HO Aegean(Roman_era_1-200AD).SG    0.561  0.0460 12.2  
2 Greek_Athens_10-20.WGA.HO Macedonia_Classical_Hellenistic 0.0531 0.0762  0.697
3 Greek_Athens_10-20.WGA.HO Montenegro_Doclea_Roman.SG      0.386  0.0413  9.36 

P 0.0707
  target                    left                         weight     se      z
  <chr>                     <chr>                         <dbl>  <dbl>  <dbl>
1 Greek_Athens_10-20.WGA.HO Aegean(Roman_era_1-200AD).SG 0.677  0.0487 13.9  
2 Greek_Athens_10-20.WGA.HO Greece_BA_Mycenaean          0.0168 0.0482  0.349
3 Greek_Athens_10-20.WGA.HO Russia_Sunghir_Medieval.SG   0.307  0.0189 16.2

P 0.0655
  target                    left                           weight     se       z
  <chr>                     <chr>                           <dbl>  <dbl>   <dbl>
1 Greek_Athens_10-20.WGA.HO Aegean(Roman_era_1-200AD).SG  0.590   0.0474 12.5   
2 Greek_Athens_10-20.WGA.HO Greece_BA_Mycenaean          -0.00186 0.0455 -0.0409
3 Greek_Athens_10-20.WGA.HO Montenegro_Doclea_Roman.SG    0.412   0.0220 18.7

P 0.0837
  target                    left                          weight     se      z
  <chr>                     <chr>                          <dbl>  <dbl>  <dbl>
1 Greek_Athens_10-20.WGA.HO Aegean(Roman_era_1-200AD).SG  0.598  0.0361 16.6  
2 Greek_Athens_10-20.WGA.HO Albania_BA_IA                -0.0145 0.0659 -0.220
3 Greek_Athens_10-20.WGA.HO Montenegro_Doclea_Roman.SG    0.417  0.0434  9.60 
---------------------------------------------------------------------------------------
P 0.341
  target    left                         weight     se     z
  <chr>     <chr>                         <dbl>  <dbl> <dbl>
1 Cretan.DG Aegean(Roman_era_1-200AD).SG  0.732 0.0338 21.7 
2 Cretan.DG Montenegro_Doclea_Roman.SG    0.268 0.0338  7.92
P 0.278
 target    left                         weight     se     z
  <chr>     <chr>                         <dbl>  <dbl> <dbl>
1 Cretan.DG Aegean(Roman_era_1-200AD).SG 0.625  0.106   5.89
2 Cretan.DG Montenegro_Doclea_Roman.SG   0.298  0.0414  7.19
3 Cretan.DG Lebanon_MBA.SG               0.0774 0.0769  1.01
```c

```
---------------------------------------------------------------------------------------
P 0.0182
  target     left                         weight     se     z
  <chr>      <chr>                         <dbl>  <dbl> <dbl>
1 Cypriot.HO Aegean(Roman_era_1-200AD).SG 0.661  0.0608 10.9 
2 Cypriot.HO Lebanon_MBA.SG               0.229  0.0537  4.25
3 Cypriot.HO Russia_Sunghir_Medieval.SG   0.0327 0.0230  1.42
4 Cypriot.HO Tajikistan_Ksirov_Kushan     0.0775 0.0269  2.88
 
P 0.00430
  target     left                         weight     se     z
  <chr>      <chr>                         <dbl>  <dbl> <dbl>
1 Cypriot.HO Aegean(Roman_era_1-200AD).SG 0.651  0.0661  9.85
2 Cypriot.HO Lebanon_MBA.SG               0.276  0.0545  5.07
3 Cypriot.HO Russia_Sunghir_Medieval.SG   0.0725 0.0181  4.00

P 0.00104
  target     left                         weight     se     z
  <chr>      <chr>                         <dbl>  <dbl> <dbl>
1 Cypriot.HO Aegean(Roman_era_1-200AD).SG  0.575 0.0950  6.05
2 Cypriot.HO Lebanon_MBA.SG                0.313 0.0691  4.53
3 Cypriot.HO Montenegro_Doclea_Roman.SG    0.112 0.0319  3.52

P 0.00389
  target     left                         weight     se     z
  <chr>      <chr>                         <dbl>  <dbl> <dbl>
1 Cypriot.HO Aegean(Roman_era_1-200AD).SG 0.621  0.0864  7.19
2 Cypriot.HO Lebanon_MBA.SG               0.248  0.0673  3.69
3 Cypriot.HO Montenegro_Doclea_Roman.SG   0.0567 0.0382  1.48
4 Cypriot.HO Tajikistan_Ksirov_Kushan     0.0741 0.0278  2.67
---------------------------------------------------------------------------------------
quick test for Turkic, didnt have Turkic on the right pops in the previous models.

P 0.00000279
  target     left                         weight      se     z
  <chr>      <chr>                         <dbl>   <dbl> <dbl>
1 Cypriot.HO Aegean(Roman_era_1-200AD).SG 0.704  0.0776   9.08
2 Cypriot.HO Lebanon_MBA.SG               0.229  0.0649   3.53
3 Cypriot.HO Russia_Sunghir_Medieval.SG   0.0481 0.0244   1.97
4 Cypriot.HO Kazakhstan_Medieval_Nomad.SG 0.0183 0.00998  1.83
---------------------------------------------------------------------------------------
P 0.152
  target                     left                         weight     se     z
  <chr>                      <chr>                         <dbl>  <dbl> <dbl>
1 Montenegro_Doclea_Roman.SG Aegean(Roman_era_1-200AD).SG  0.197 0.0423  4.65
2 Montenegro_Doclea_Roman.SG Croatia_EIA                   0.118 0.0677  1.75
3 Montenegro_Doclea_Roman.SG Russia_Sunghir_Medieval.SG    0.685 0.0496 13.8 
---------------------------------------------------------------------------------------
P 0.149
  target      left                       weight       se       z
  <chr>       <chr>                       <dbl>    <dbl>   <dbl>
1 Croatian.HO Montenegro_Doclea_Roman.SG      1 1.14e-13 8.79e12
---------------------------------------------------------------------------------------
P 0.0000798
  target                     left                       weight     se     z
  <chr>                      <chr>                       <dbl>  <dbl> <dbl>
1 Montenegro_Doclea_Roman.SG Croatia_EIA                 0.233 0.0579  4.02
2 Montenegro_Doclea_Roman.SG Russia_Sunghir_Medieval.SG  0.767 0.0579 13.2 
---------------------------------------------------------------------------------------
P 0.0121
  target      left                         weight     se      z
  <chr>       <chr>                         <dbl>  <dbl>  <dbl>
1 Albanian.HO Aegean(Roman_era_1-200AD).SG 0.536  0.0464 11.5  
2 Albanian.HO Albania_BA_IA                0.0594 0.0716  0.829
3 Albanian.HO Russia_Sunghir_Medieval.SG   0.405  0.0374 10.8  
---------------------------------------------------------------------------------------
P 0.000771
  target      left                          weight     se      z
  <chr>       <chr>                          <dbl>  <dbl>  <dbl>
1 Albanian.HO Aegean(Roman_era_1-200AD).SG  0.450  0.0411 11.0  
2 Albanian.HO Albania_BA_IA                -0.0416 0.0710 -0.586
3 Albanian.HO Montenegro_Doclea_Roman.SG    0.591  0.0465 12.7
---------------------------------------------------------------------------------------
P 0.00204
  target      left                         weight     se     z
  <chr>       <chr>                         <dbl>  <dbl> <dbl>
1 Albanian.HO Aegean(Roman_era_1-200AD).SG  0.432 0.0252  17.2
2 Albanian.HO Montenegro_Doclea_Roman.SG    0.568 0.0252  22.6
---------------------------------------------------------------------------------------
P 0.118
  target       left                         weight     se     z
  <chr>        <chr>                         <dbl>  <dbl> <dbl>
1 Bulgarian.HO Aegean(Roman_era_1-200AD).SG  0.449 0.0227  19.8
2 Bulgarian.HO Russia_Sunghir_Medieval.SG    0.551 0.0227  24.2

P 0.0473
  target       left                           weight     se      z
  <chr>        <chr>                           <dbl>  <dbl>  <dbl>
1 Bulgarian.HO Aegean(Roman_era_1-200AD).SG  0.453   0.0501  9.03 
2 Bulgarian.HO Bulgaria_EIA                 -0.00536 0.0530 -0.101
3 Bulgarian.HO Russia_Sunghir_Medieval.SG    0.553   0.0243 22.7 

P 0.00666
  target       left                         weight     se     z
  <chr>        <chr>                         <dbl>  <dbl> <dbl>
1 Bulgarian.HO Aegean(Roman_era_1-200AD).SG  0.256 0.0244  10.5
2 Bulgarian.HO Montenegro_Doclea_Roman.SG    0.744 0.0244  30.5

P 0.00106
  target       left                          weight     se     z
  <chr>        <chr>                          <dbl>  <dbl> <dbl>
1 Bulgarian.HO Aegean(Roman_era_1-200AD).SG  0.317  0.0447  7.09
2 Bulgarian.HO Bulgaria_EIA                 -0.0727 0.0462 -1.57
3 Bulgarian.HO Montenegro_Doclea_Roman.SG    0.755  0.0262 28.8 

P 0.0473
  target       left                           weight     se      z
  <chr>        <chr>                           <dbl>  <dbl>  <dbl>
1 Bulgarian.HO Aegean(Roman_era_1-200AD).SG  0.453   0.0501  9.03 
2 Bulgarian.HO Bulgaria_EIA                 -0.00536 0.0530 -0.101
3 Bulgarian.HO Russia_Sunghir_Medieval.SG    0.553   0.0243 22.7 


DNA samples from the Version v54.1.p1 Harvard dataset: https://reichdata.hms.harvard.edu/pub/datasets/amh_repo/curated_releases/

No comments:

Post a Comment