Comparing Ancient Greek populations to modern Greeks and Italians

Jovialis said:
But that's not to say it did not impact southern Italy at all. Maybe just less, or more in some areas than others?

I think we do see the legacy of these broader Greek-speaking world migrations more in modern Greeks actually, because most get Minoan_Zakros (about 30% Levant_PPN admixed if I recall), as well we more Kura-Araxes in modern Greeks. Though that may just be how the model makes up for the Corded ware brought in later in the MA. This would need to be resolved by ancIBD.

jdn6qSJ.png


As for where the Greek colonists originated from, Achaean was widely spoken throughout much of the south. That with very little doubt came straight from Northern Peloponnesus.
Click to expand...

Forgot to post the map:

1QR7Iw2.png
 
Just trying some qpAdm models for Greeks, Italians, Levantines.

Keep in mind that Levantines, even in studies try to inflate their Levant Neolithic by not including Anatolian Neolithic as a proxy. Once you make proper models such as these below, you see the reality.

right = c('Ethiopia_4500BP.DG', 'Russia_MA1_HG.SG', 'Morocco_Iberomaurusian', 'Turkey_Epipaleolithic', 'Turkey_Boncuklu_N', 'Georgia_Satsurblia', 'Sidelkino_noUDG_EHG', 'Bichon_WHG')


left = c('Turkey_N', 'Israel_Natufian', 'Iran_TepeAbdulHosein_N', 'Russia_Samara_EBA_Yamnaya', 'Serbia_IronGates_Mesolithic')


target = c('Greek_Cypriot')


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


Code: 
```c
P 0.211
  target              left                        weight     se     z
  <chr>               <chr>                        <dbl>  <dbl> <dbl>
1 Greece_BA_Mycenaean Turkey_N                    0.679  0.0177 38.3 
2 Greece_BA_Mycenaean Israel_Natufian             0.0337 0.0180  1.88
3 Greece_BA_Mycenaean Iran_TepeAbdulHosein_N      0.101  0.0251  4.01
4 Greece_BA_Mycenaean Russia_Samara_EBA_Yamnaya   0.171  0.0286  6.00
5 Greece_BA_Mycenaean Serbia_IronGates_Mesolithic 0.0152 0.0127  1.20


P 0.282
  target        left                        weight     se     z
  <chr>         <chr>                        <dbl>  <dbl> <dbl>
1 Greek_Cypriot Turkey_N                    0.491  0.0201 24.4 
2 Greek_Cypriot Israel_Natufian             0.0488 0.0224  2.18
3 Greek_Cypriot Iran_TepeAbdulHosein_N      0.325  0.0334  9.73
4 Greek_Cypriot Russia_Samara_EBA_Yamnaya   0.107  0.0363  2.96
5 Greek_Cypriot Serbia_IronGates_Mesolithic 0.0280 0.0155  1.80


P 0.185
  target       left                      weight     se      z
  <chr>        <chr>                      <dbl>  <dbl>  <dbl>
1 Greek_Athens Turkey_N                  0.488  0.0142 34.3  
2 Greek_Athens Israel_Natufian           0.0151 0.0156  0.965
3 Greek_Athens Iran_TepeAbdulHosein_N    0.217  0.0223  9.74 
4 Greek_Athens Russia_Samara_EBA_Yamnaya 0.203  0.0254  7.99 
5 Greek_Athens Lithuania_EMN_Narva       0.0772 0.0111  6.98 
```
```c
P 0.434
  target        left                       weight      se     z
  <chr>         <chr>                       <dbl>   <dbl> <dbl>
1 Italian_North Turkey_N                   0.534  0.0140  38.2 
2 Italian_North Israel_Natufian           -0.0160 0.0150  -1.07
3 Italian_North Iran_TepeAbdulHosein_N     0.131  0.0205   6.38
4 Italian_North Russia_Samara_EBA_Yamnaya  0.265  0.0205  13.0 
5 Italian_North Loschbour_WHG              0.0854 0.00697 12.3


P 0.430
  target        left                      weight      se     z
  <chr>         <chr>                      <dbl>   <dbl> <dbl>
1 Italian_North Turkey_N                  0.525  0.0101  52.1 
2 Italian_North Iran_TepeAbdulHosein_N    0.119  0.0178   6.69
3 Italian_North Russia_Samara_EBA_Yamnaya 0.272  0.0199  13.7 
4 Italian_North Loschbour_WHG             0.0847 0.00691 12.3 


PPNB in right.

Natufian inflated, North African deflated, a tiny bit of North African being deflated due to noise will increase Natufian. My guess is that South Italy should score around 2% based on what Athens and Cyprus scores. 
P 0.101
  target        left                       weight      se      z
  <chr>         <chr>                       <dbl>   <dbl>  <dbl>
1 Italian_South Turkey_N                  0.479   0.0307  15.6  
2 Italian_South Israel_Natufian           0.0409  0.0384   1.06 
3 Italian_South Iran_TepeAbdulHosein_N    0.184   0.0313   5.89 
4 Italian_South Russia_Samara_EBA_Yamnaya 0.243   0.0312   7.81 
5 Italian_South Loschbour_WHG             0.0450  0.00990  4.54 
6 Italian_South Morocco_EN                0.00716 0.00843  0.849

I'm not sure if the Levantine steppe levels are real or if they actually need excess Euro HG stuff. If i add iron gates to the right for Lebanese Muslims, the model fails, which is a way of qpAdm telling me that its needed.

Code: 
```cP 0.842
  target      left                   weight     se     z
  <chr>       <chr>                   <dbl>  <dbl> <dbl>
1 Lebanon_MBA Turkey_N                0.421 0.0230 18.3 
2 Lebanon_MBA Israel_Natufian         0.193 0.0261  7.39
3 Lebanon_MBA Iran_TepeAbdulHosein_N  0.386 0.0189 20.4 


P 0.0542
  target             left                      weight     se     z
  <chr>              <chr>                      <dbl>  <dbl> <dbl>
1 Lebanese_Christian Turkey_N                  0.409  0.0195 20.9 
2 Lebanese_Christian Israel_Natufian           0.153  0.0218  7.01
3 Lebanese_Christian Iran_TepeAbdulHosein_N    0.343  0.0290 11.9 
4 Lebanese_Christian Russia_Samara_EBA_Yamnaya 0.0951 0.0236  4.03
```


P 0.265
  target             left                        weight     se      z
  <chr>              <chr>                        <dbl>  <dbl>  <dbl>
1 Lebanese_Christian Turkey_N                    0.404  0.0196 20.6  
2 Lebanese_Christian Israel_Natufian             0.149  0.0215  6.91 
3 Lebanese_Christian Iran_TepeAbdulHosein_N      0.381  0.0345 11.0  
4 Lebanese_Christian Russia_Samara_EBA_Yamnaya   0.0315 0.0385  0.819
5 Lebanese_Christian Serbia_IronGates_Mesolithic 0.0345 0.0155  2.23 


Iron gates in right
P 0.0383
  target             left                      weight     se     z
  <chr>              <chr>                      <dbl>  <dbl> <dbl>
1 Lebanese_Christian Turkey_N                  0.411  0.0197 20.9 
2 Lebanese_Christian Israel_Natufian           0.148  0.0222  6.67
3 Lebanese_Christian Iran_TepeAbdulHosein_N    0.368  0.0239 15.4 
4 Lebanese_Christian Russia_Samara_EBA_Yamnaya 0.0732 0.0182  4.0


P 0.0957
  <chr>           <chr>                      <dbl>   <dbl> <dbl>
1 Lebanese_Muslim Turkey_N                  0.404  0.0218  18.5 
2 Lebanese_Muslim Israel_Natufian           0.106  0.0244   4.36
3 Lebanese_Muslim Iran_TepeAbdulHosein_N    0.343  0.0323  10.6 
4 Lebanese_Muslim Russia_Samara_EBA_Yamnaya 0.114  0.0233   4.90
5 Lebanese_Muslim Dinka.DG                  0.0318 0.00852  3.74


P 0.179
  target          left                        weight      se     z
  <chr>           <chr>                        <dbl>   <dbl> <dbl>
1 Lebanese_Muslim Turkey_N                    0.399  0.0223  17.9 
2 Lebanese_Muslim Israel_Natufian             0.107  0.0248   4.30
3 Lebanese_Muslim Iran_TepeAbdulHosein_N      0.373  0.0377   9.91
4 Lebanese_Muslim Russia_Samara_EBA_Yamnaya   0.0640 0.0362   1.77
5 Lebanese_Muslim Serbia_IronGates_Mesolithic 0.0273 0.0141   1.94
6 Lebanese_Muslim Dinka.DG                    0.0299 0.00896  3.33


P 0.290
  target    left                        weight      se     z
  <chr>     <chr>                        <dbl>   <dbl> <dbl>
1 Jordanian Turkey_N                    0.314  0.0237  13.2 
2 Jordanian Israel_Natufian             0.193  0.0255   7.57
3 Jordanian Iran_TepeAbdulHosein_N      0.297  0.0411   7.23
4 Jordanian Russia_Samara_EBA_Yamnaya   0.0987 0.0406   2.43
5 Jordanian Serbia_IronGates_Mesolithic 0.0311 0.0157   1.98
6 Jordanian Dinka.DG                    0.0657 0.00962  6.83


P 0.110
  target    left                      weight      se     z
  <chr>     <chr>                      <dbl>   <dbl> <dbl>
1 Jordanian Turkey_N                  0.323  0.0231  14.0 
2 Jordanian Israel_Natufian           0.192  0.0253   7.60
3 Jordanian Iran_TepeAbdulHosein_N    0.256  0.0343   7.46
4 Jordanian Russia_Samara_EBA_Yamnaya 0.159  0.0248   6.42
5 Jordanian Dinka.DG                  0.0690 0.00936  7.37
 
Jovialis said:
There's been a lot of South Eastern European input into the Middle East as well, particularly from Greeks. I think it is likely some of the EHG may come from that.
Click to expand...

I meant the excess Euro HG %, they score some excess Euro hg, not just steppe.

What do you think of the models i sent for Greeks and Italians
 
TBH, the Anatolian_N looks a bit low for South Italians. Are you using Barcin_N? I saw a map of G25 results going around and it looks almost uniform of about 60% from North to South. I would post it, bit I'm on a cruise off the coast of Akita at the moment on my cellphone.
 
Jovialis said:
TBH, the Anatolian_N looks a bit low for South Italians. Are you using Barcin_N? I saw a map of G25 results going around and it looks almost uniform of about 60% from North to South. I would post it, bit I'm on a cruise off the coast of Akita at the moment on my cellphone.
Click to expand...

On the link you sent, they used the old Cypriot average that had Maronite samples as Greek Cypriot so ANF was deflated, actual Greek Cypriots have similar ANF levels to South Italians but both actually score less on qpAdm. If qpAdm shows something different than G25 then G25 is wrong.

I also think that ANF on South Italians should probably be 2% higher on that qpAdm model, its just Standard Errors, not a big difference. Always keep in mind the Standard Errors. When i run many more models with a bit different references and proxies we will get to see a pattern forming of the admixture clines.

What i expect is 50% ANF on South Italians and 47% on Cypriots.

Yes im using Barcin_N. G25 is inaccurate on Neolithic models. It inflates and deflates ANF on populations depending on the PCA position. It is a just PCA meant to tell apart modern populations so the older the model the more inaccurate, just a bit of drift will mess up since it has to place that population on another PCA position. Even without drift it will still mess up because Neolithic is just too old for a PCA meant to differentiate modern populations.

For example if you make an average of Sardinian and Cypriot on G25, the PPNB levels almost dissappear unproportionally instead of being halved. That happens because a PCA meant for modern people simply can't model accurately ancestry as far as the Neolithic.

The misuse of G25 with Neolithic models did a lot of damage to population genetics. G25 good on proximal models, try to keep it Bronze age and after.

I was also chatting about DNA stuff on a cruise ship lol.
 
Idontknowwhatimdoing said:
G25 good on proximal models, try to keep it Bronze age and after.

I was also chatting about DNA stuff on a cruise ship lol.
Click to expand...

I find the same is true for Dodecad K12b.
At any rate, I don't know how to run qpAdm (yet) but I think I'm going to get cracking on that after my vacation. Today's a sea day, thus my presence here online atm.
 
Francesco said:
The results are a bit different from the other model you posted here: https://www.eupedia.com/forum/threa...talians/page33?p=668591&viewfull=1#post668591

What do you think is the most accurate model between the two?
Click to expand...

In this one i used Anatolian Neolithic instead of Anatolian HG as a proxy. I also changed the Iran N proxy and used the higher coverage Mota sample as outgroup.

In my old models i did Anatolian HG vs Natufian and Anatolian Neolithic vs Levant Neolithic. Now i did Anatolian Neolithic vs Natufian.
 
Jovialis said:
I find the same is true for Dodecad K12b.
At any rate, I don't know how to run qpAdm (yet) but I think I'm going to get cracking on that after my vacation. Today's a sea day, thus my presence here online atm.
Click to expand...

I can guide you.
 
Idontknowwhatimdoing said:
I told him to share that map, it is G25 distance and not admixture. Basically these numbers below.
[FONT=&quot]
xI77dLt.png

[/FONT][FONT=&quot]
oj1lVuA.png

[/FONT][FONT=&quot]
y8ExLqR.png
[/FONT]
Click to expand...
A map demonstration of admixture in qpADM would be interesting. But perhaps Greece_N would be even better.

Yeah that would be cool. I'll probably post questions in the thread.
 
There's a study that demonstrates AI can reduce the learning curve of new hires by 6 months. I'm hoping that will also aid me in this endeavor.

Chatgpt: qpAdm is a tool used in population genetics to model a target population as a mixture of two or more source populations. It's often used to model modern populations using ancient DNA. Here's a general step-by-step guide on how to run qpAdm:

1. **Prepare the Input Files**: The first step is to prepare the input files. You will need a parameter file, a genotype file, and a SNP file. The parameter file specifies the details of the analysis, the genotype file contains the genetic data, and the SNP file contains information about the SNPs.

2. **Specify the Populations**: In the parameter file, specify the target population and the source populations. The target population is the one you are trying to model, and the source populations are the ones you are using to model the target.

3. **Choose the Outgroup**: Choose an outgroup population. This should be a population that is not closely related to any of the source or target populations.

4. **Run qpAdm**: Once you have prepared the input files and specified the populations, you can run qpAdm. The command to run qpAdm is typically something like this: `qpAdm -p parameterfile.par`.

5. **Interpret the Results**: qpAdm will output a number of statistics, including the proportion of each source population in the target population and a p-value for the model. If the p-value is high, this suggests that the model is a good fit for the data.

Remember that the specifics of how to run qpAdm can vary depending on the details of your analysis and the format of your data. Always refer to the qpAdm documentation and any relevant literature to ensure you are using the tool correctly.

If you need a visual guide or a more detailed explanation, I can create a diagram or search for scientific literature that explains the process in more detail.
 
Jovialis said:
A map demonstration of admixture in qpADM would be interesting. But perhaps Greece_N would be even better.

Yeah that would be cool. I'll probably post questions in the thread.
Click to expand...

Greece_N are samples from multiple periods, the older the more ANF and less Iran N.
 
Idontknowwhatimdoing said:
Greece_N are samples from multiple periods, the older the more ANF and less Iran N.
Click to expand...
That's correct, like I said it would be interesting to see a map using that as an admixture component. I don't think it would be unreasonable considering academic studies use "mixed" components as well.
"Steppe/Yamna" usually comes as a package "EHG+CHG" how could one decern excess non-steppe related CHG when only using EHG and CHG? But also even "Natufian" is mixed, as well as Anatolia Hunter-gatherers. Frankly, I'm not entirely convinced that is an appropriate method. Nevertheless, I can respect your insights considering you have knowledge with qpADM as a tool.
That being said, I still think my approach to the topic is indeed viable. I hope you can respect that.
 
You must log in or register to reply here.

This thread has been viewed 134067 times.

Back
Top