Mega QAPDM Models thread

[Opie]

Here's a model with insanely bloated Syria_MBA. I couldn't find Sidon sample and used Syria instead.

[Gannicus]

A two-way model I've seen on IllustrativeDNA in the bronze age section of deep ancestry:



G25:
Target: Gannicus_MergedFile_officialDavidski_scaled
Distance: 0.0352% / 0.03515623
84.1 Czech_EBA_Unetice
15.9 Greece_Manika_Helladic_EBA.SG

[Grace O'Malley]

I'm just looking at this in dribs and drabs and trying to familiarise myself with it's use. I've now realised you don't have to use single samples and I've done this FST distance chart. I'm presuming the negative is a closer distance? I selected the AADR dataset for the populations. There isn't a lot of modern populations to choose from.

Orcadians showing obvious drift which is why they are more distant.



I used the Human Origins dataset for this one which has more modern populations. I understand the distances better now. I think when I asked previously I didn't notice that the top pops where negative.

[Grace O'Malley]

This is my best model using moderns so far others were coming up False.



This one is not as close but more realistic than French.





I'll try other models and will try some ancient ones that are in this thread.

[Opie]

Here's another model that I used



Sources 6 / 10
Turkey_Marmara_Barcin_N.AG


Iran_Wezmeh_N.SG


Georgia_Kotias_Mesolithic.SG


Russia_YuzhniyOleniyOstrov_Mesolithic.AG


Israel_Natufian.AG


Mongolia_North_N.AG




References 13 / 30
Mbuti.DG


Austria_Gravettian.AG


Russia_MA1_UP.SG


Russia_LakeBaikal_N.SG


Russia_LenaRiver_N.SG


Turkey_Central_Pinarbasi_Epipaleolithic.AG


Morocco_EN.WGC.SG


Turkey_Central_Boncuklu_PPN.WGC.SG


Russia_Sidelkino_HG.SG


China_Xinjiang_Xiaohe_BA.AG


Georgia_Satsurblia_LateUP.SG


Iran_TepeAbdulHosein_N.SG


India_GreatAndaman_100BP.SG

[Gannicus]

I'm just looking at this in dribs and drabs and trying to familiarise myself with it's use. I've now realised you don't have to use single samples and I've done this FST distance chart. I'm presuming the negative is a closer distance? I selected the AADR dataset for the populations. There isn't a lot of modern populations to choose from.

Orcadians showing obvious drift which is why they are more distant.



I used the Human Origins dataset for this one which has more modern populations. I understand the distances better now. I think when I asked previously I didn't notice that the top pops where negative.

From what I understand there is really no such thing as negative. It means 0 and very close.

[Opie]

This model is where I was experimenting with Serednii Stih/Sredny Stog.

Attachment 142635
Attachment 142636

It seems you really have Natufian admixture despite the fact that your ancestry is NW Euro.

[Opie]

Here's another model that I used



Sources 6 / 10
Turkey_Marmara_Barcin_N.AG


Iran_Wezmeh_N.SG


Georgia_Kotias_Mesolithic.SG


Russia_YuzhniyOleniyOstrov_Mesolithic.AG


Israel_Natufian.AG


Mongolia_North_N.AG




References 13 / 30
Mbuti.DG


Austria_Gravettian.AG


Russia_MA1_UP.SG


Russia_LakeBaikal_N.SG


Russia_LenaRiver_N.SG


Turkey_Central_Pinarbasi_Epipaleolithic.AG


Morocco_EN.WGC.SG


Turkey_Central_Boncuklu_PPN.WGC.SG


Russia_Sidelkino_HG.SG


China_Xinjiang_Xiaohe_BA.AG


Georgia_Satsurblia_LateUP.SG


Iran_TepeAbdulHosein_N.SG


India_GreatAndaman_100BP.SG

5.05% Natufian with 5.93% error margin this component irrelevant for me.

[Opie]

Here's another model that I used this one seems better than the previous.



Sources 5 / 10
Turkey_Marmara_Barcin_N.AG


Georgia_Kotias_Mesolithic.SG


Iran_GanjDareh_N.AG


Russia_Murzikha_Eneolithic_Sidelkino.AG


Mongolia_North_N.AG




References 13 / 30
Mbuti.DG


Austria_Gravettian.AG


Russia_MA1_UP.SG


Russia_LakeBaikal_N.SG


Russia_LenaRiver_N.SG


Turkey_Central_Pinarbasi_Epipaleolithic.AG


Morocco_EN.WGC.SG


Turkey_Central_Boncuklu_PPN.WGC.SG


Russia_Sidelkino_HG.SG


China_Xinjiang_Xiaohe_BA.AG


Georgia_Satsurblia_LateUP.SG


Iran_TepeAbdulHosein_N.SG


India_GreatAndaman_100BP.SG

[Opie]

5.05% Natufian with 5.93% error margin this component irrelevant for me.

Since barcin might mask some Natufian. I used Aktopraklik instead and the margion of error is lower than the percentage.