15 Thousand Years Ago by Tomenable / Peterski

Quote:

---

Originally Posted by Vessna

Target: Vessna_scaled
Distance: 6.6341% / 0.06634091
35.0 Turkey_Barcin_LN.SG
34.8 Russia_Samara_EBA_Yamnaya
15.4 Russia_Karelia_HG
14.0 Latvia_HG.SG
0.8 Hungary_EN_HG_Koros

Target: husband_scaled
Distance: 5.5742% / 0.05574151
31.2 Russia_Samara_EBA_Yamnaya
29.2 Turkey_Barcin_LN.SG
20.4 Russia_Karelia_HG
13.4 Latvia_HG.SG
3.0 Japan_HG_Jomon
2.2 Hungary_EN_HG_Koros
0.6 Russia_Ekven_OldBeringSea

---

I made a Ghost G25 sample (75% Modern Dutch, 25% Modern Finnish_East): basic 3-way result for Yamnaya for this Ghost: and the Ghost sample is compared to my result below-- not far apart. But this new member is getting over 59% Yamnaya on this calc.

Dutch.FinnEast.Ghost,0.1270445,0.12032825,0.068800 25000000001,0.054017499999999996,0.0374805,0.01751 6999999999998,0.005939999999999999,0.0089335,0.004 3035,-0.0062045,-0.002382,0.00134525,-0.0033282500000000005,-0.005986,0.0123545,0.0064427500000000006,-0.0021349999999999997,0.0008992500000000001,0.0025 905000000000004,0.0036172500000000002,0.0057977499 99999999,0.0028215000000000002,0.00029549999999999 997,0.012420249999999999,0.000025499999999999915


Target: Dutch.FinnEast.Ghost
Distance: 4.0090% / 0.04009013
53.0 Russia_Samara_EBA_Yamnaya
33.6 Turkey_Barcin_LN.SG
13.4 France_Rochedane

Target: SC_Official_AVE_Scaled
Distance: 3.8097% / 0.03809678
53.4 Russia_Samara_EBA_Yamnaya
35.4 Turkey_Barcin_LN.SG
11.2 France_Rochedane

Quote:

---

Originally Posted by Vessna

Makes no sense for you to have more Yamnaya than a Northern Russian.

---

I don't. It's interesting that I get 53.4% on the basic 3-way, but also 53.4% when adding in multiple other HG and Farmer samples. Most people's Yamnaya scores will change when adding in others.

Target Russia_Samara_EBA_Yamnaya
Russian_Leshukonsky 68.8
Russian_Pinezhsky 64.6
Russian_Pinega 63
Finnish_East 62.4
Karelian 61.8
Finnish_North 61.6
Russian_Krasnoborsky 61.4
Russian_Kostroma 61
Finnish_Central 61
Finnish_Southeast 59.4
Russian_Tver 58.2
Russian_Yaroslavl 58
Finnish_Southwest 58
Russian_Ryazan 57.4
Russian_Kursk 56.2
Russian_Orel 56
Russian_Pskov 55.8
Russian_Kaluga 54.8
Belarusian 54.4
Russian_Smolensk 54.2
Russian_Voronez 54.2
Russian_Belgorod 53.6
SC_Official_AVE_Scaled 53.4
Shetlandic 53.2
Dutch.FinnEast.Ghost 53
Swedish 52.8
Icelandic 52.2
Norwegian 51.8
Irish 51.6
Danish 51.2
Scottish 50.6
Dutch 49.8
English 49
Welsh 48.6
English_Cornwall 48

Quote:

---

Originally Posted by Asega

Balto-Slavs (including you) suffer from Balto-Slavic drift (genetic drift) on G25, which is why you latch onto Latvian HG's.

Davidski from Eurogenes has a blogpost explaining the phenomenon:

https://eurogenes.blogspot.com/2021/...st-vs.html?m=1

---

Thanks! I am aware of the drift just wasn’t sure how it would affect calc results.

Found this article from Davidski.
https://eurogenes.blogspot.com/2012/...s-all.html?m=1

Interestingly, when I asked ChatGPT that same question several weeks ago, I got a similar answer :)

Quote:

---

Originally Posted by ScandinavianCelt

I don't.

Target Russia_Samara_EBA_Yamnaya
Russian_Leshukonsky 68.8
Russian_Pinezhsky 64.6
Russian_Pinezhsky 64.6
Russian_Pinega 63
Finnish_East 62.4
Karelian 61.8
Finnish_North 61.6
Russian_Krasnoborsky 61.4
Russian_Kostroma 61
Finnish_Central 61
Finnish_Southeast 59.4
Russian_Tver 58.2
Russian_Yaroslavl 58
Finnish_Southwest 58
Russian_Ryazan 57.4
Russian_Kursk 56.2
Russian_Orel 56
Russian_Pskov 55.8
Russian_Kaluga 54.8
Belarusian 54.4
Russian_Smolensk 54.2
Russian_Voronez 54.2
Russian_Belgorod 53.6
SC_Official_AVE_Scaled 53.4
Shetlandic 53.2
Dutch.FinnEast.Ghost 53
Swedish 52.8
Icelandic 52.2
Norwegian 51.8
Irish 51.6
Danish 51.2
Scottish 50.6
Dutch 49.8
English 49
Welsh 48.6
English_Cornwall 48

---

Well that’s not what I got for him on your calc. His closest population is Russian Krasnoborsky or Kostroma depending on the calc.
Edit: never mind, I missed your previous post explaining why you added more HG samples.

Quote:

---

Originally Posted by ScandinavianCelt

I don't. It's interesting that I get 53.4% on the basic 3-way, but also 53.4% when adding in multiple other HG and Farmer samples. Most people's Yamnaya scores will change when adding in others.

Target Russia_Samara_EBA_Yamnaya
Russian_Leshukonsky 68.8
Russian_Pinezhsky 64.6
Russian_Pinezhsky 64.6
Russian_Pinega 63
Finnish_East 62.4
Karelian 61.8
Finnish_North 61.6
Russian_Krasnoborsky 61.4
Russian_Kostroma 61
Finnish_Central 61
Finnish_Southeast 59.4
Russian_Tver 58.2
Russian_Yaroslavl 58
Finnish_Southwest 58
Russian_Ryazan 57.4
Russian_Kursk 56.2
Russian_Orel 56
Russian_Pskov 55.8
Russian_Kaluga 54.8
Belarusian 54.4
Russian_Smolensk 54.2
Russian_Voronez 54.2
Russian_Belgorod 53.6
SC_Official_AVE_Scaled 53.4
Shetlandic 53.2
Dutch.FinnEast.Ghost 53
Swedish 52.8
Icelandic 52.2
Norwegian 51.8
Irish 51.6
Danish 51.2
Scottish 50.6
Dutch 49.8
English 49
Welsh 48.6
English_Cornwall 48

---

Looks like you modeled them without the East Asian component

Target: scaled
Distance: 4.6820% / 0.04681988
46.2 Anatolia_15KYA:ZBC_IPB001
29.6 Eastern_Europe_15KYA:PES001
14.8 Caucasus_15KYA:KK1
5.4 Iran_15KYA:WC1
4.0 Levant_15KYA:Natufian


Target: scaled
Distance: 3.2779% / 0.03277912
49.4 Russia_Samara_EBA_Yamnaya
46.0 Turkey_Barcin_LN.SG
4.6 France_Rochedane

Quote:

---

Originally Posted by Vessna

Thanks! I am aware of the drift just wasn’t sure how it would affect calc results.

Found this article from Davidski.
https://eurogenes.blogspot.com/2012/...s-all.html?m=1

Interestingly, when I asked ChatGPT that same question several weeks ago, I got a similar answer :)

---

It's best to omit Latvian HG's all together because no one in Europe has any real ancestry from them. You can be perfectly modelled without it.

Quote:

---

Originally Posted by Asega

I'm curious to know what you think the variables are at stake.

---

Accuracy of individual samples in Source.
Accuracy of combined samples in Source.
Accuracy of Target coordinates.
Accuracy of Target raw-data.
Individual inheritance within a nationality/ethnicity (example: the range for ALL Dutch samples on the 3-way: 43.6%-54.6%)


etc etc

Quote:

---

Originally Posted by ScandinavianCelt

Accuracy of individual samples in Source.
Accuracy of combined samples in Source.
Accuracy of Target coordinates.
Accuracy of Target raw-data.
Individual inheritance within a nationality/ethnicity (example: the range for ALL Dutch samples on the 3-way: 43.6%-54.6%)


etc etc

---

There is quite some sub-strucure within the population here which explains the variance. This has to do with differences in ancient ancestry between regions in NL.

Some parts of NL are mostly Celtic in ancestry. Which explains why some Dutch are poorer in Steppe ancestry relative to say someone with North Sea Germanic/Frisian ancestry (like myself).

Quote:

---

Originally Posted by Asega

It's best to omit Latvian HG's all together because no one in Europe has any real ancestry from them. You can be perfectly modelled without it.

---

Target: husband_scaled
Distance: 6.7195% / 0.06719516
53.8 Russia_Samara_EBA_Yamnaya
23.6 Turkey_Barcin_LN.SG
17.6 France_Rochedane
5.0 Russia_Ekven_OldBeringSea

Target: Vessna_scaled
Distance: 7.5849% / 0.07584930
55.4 Russia_Samara_EBA_Yamnaya
30.0 Turkey_Barcin_LN.SG
14.6 France_Rochedane