Eurogenes Biogeographic Ancestry Project

[Argang]

Here's the other Saami, and her K13 results:

1 Baltic 44.78
2 North_Atlantic 25.28
3 Siberian 22.22
4 Amerindian 3.13
5 South_Asian 1.46
6 East_Asian 1.25
7 Sub-Saharan 0.95
8 Oceanian 0.93

K15:

1 Eastern_Euro 27.74
2 North_Sea 24.11
3 Siberian 20.72
4 Baltic 14.6
5 Atlantic 7.58
6 Amerindian 3
7 Southeast_Asian 0.97
8 Oceanian 0.81
9 Sub-Saharan 0.32
10 South_Asian 0.14

Her EEF-WHG-ANE results:

EEF 29,01530701
WHG 50,75505548
ANE 20,22963751

and for the other saami

EEF 29,37181261
WHG 50,57478162
ANE 20,05340577

High Siberian distorts the results a bit, but likely negligible regarding ANE. In the f4 test included in that Lazaridis paper, the single Saami sample they tested shared more with Mal'ta than anyone else did, even though it was also more eastern than anyone else.

[ZephyrousMandaru]

Eurogenes K13 results of a Norwegian Saami gedmatch relative to me:

1 Baltic 44.35
2 North_Atlantic 24.18
3 Siberian 22.75
4 South_Asian 2.77
5 Amerindian 2.44
6 West_Med 1.51
7 Oceanian 1.04
8 West_Asian 0.53
9 Sub-Saharan 0.42

EUTest K15:

1 Eastern_Euro 26.01
2 North_Sea 22.27
3 Siberian 21.43
4 Baltic 15.45
5 Atlantic 10.26
6 Amerindian 2.18
7 South_Asian 1.57
8 Oceanian 0.82

Something tells me the bulk of that Siberian is actually ANE.

[Argang]

Something tells me the bulk of that Siberian is actually ANE.

Could be, we have yet to see a Saami - MA-1 comparison.

Looking at the lack of West Asian and mediterranean in those results, it's safe to say they'd cluster noticeably to the left of Chuvashes, Tatars, Maris etc in this PCA like this.

[Argang]

There's a new experiment going on, goal seems to be to calculate "real" ANE, ENA, Middle Eastern and Sub-Saharan ancestry by fitting populations between Mal'ta, Han Chinese, Bedouins and Yoruba. I'll add a reminder that the accuracy of these results is based on the assumption that Bedouins and Han do not have any ANE. Results Polako posted so far:

Sample: ANE (MA-1), ENA (Han), ME (Bedouin), SSA (Yoruba)
Abhkasian: 0.195, 0.055, 0.75, 0
Abhkasian: 0.176, 0.055, 0.768, 0
Chechen: 0.249, 0.089, 0.662, 0
Chechen: 0.24, 0.094, 0.665, 0
Cyprian: 0.141, 0.026, 0.833, 0
Cyprian: 0.107, 0.048, 0.845, 0
Lezgin: 0.279, 0.066, 0.655, 0
Lezgin: 0.283, 0.055, 0.662, 0
Afghan_Pashtun: 0.288, 0.19, 0.521, 0

https://docs.google.com/spreadsheet/...&rm=full#gid=0

The Bedouin sample in the K13 test is actually 6% Northeast African (peaks in Gumuz and Ari cultivators) and 3% SSA (peaks in Yoruba), so one would need to be noticeably SSA-admixed for it to show in this kind of test.

[Harkonnen]

So here's the recent update.

http://eurogenes.blogspot.com.au/201...ne-levels.html

What I found especially noteworthy is the relatively low levels of actual ANE ancestry among Siberians, yet we know that they share rather high with Mal'ta. This must mean that their ENA-like ancestry contributes somewhat to this similarity (or at least does not seem to pull them away much) contra some other ancestries, namely South Asian, but perhaps difficult to compare with the mideastern ones -> think hat.

Similarly WHG ancestry ads to the affinity to ANE.

[Argang]

So here's the recent update.

http://eurogenes.blogspot.com.au/201...ne-levels.html

What I found especially noteworthy is the relatively low levels of actual ANE ancestry among Siberians, yet we know that they share rather high with Mal'ta. This must mean that their ENA-like ancestry contributes somewhat to this similarity (or at least does not seem to pull them away much) contra some other ancestries, namely South Asian, but perhaps difficult to compare with the mideastern ones -> think hat.

Similarly WHG ancestry ads to the affinity to ANE.

It highlights the importance of picking optimal reference populations for this kind of test, or perhaps even lack of optimal reference populations for this kind of test. If there is ANE ancestry in general Han Chinese population - they do show higher affinity to MA-1 than Bedouins as we know from the peer-reviewed study that introduced the genome - the result will be every population with ANE getting also some ENA (Han). Even Saudis and Palestinians, which can be seen in the averages. Additionally some East Asians closest to Han average will get no ANE because the Han covers it all.

She might have been a better ENA reference population than Han looking at the f4 test in the initial MA-1 study, although not by much.

Some North Han are getting small ANE in this test, which fits as they must be more "northern" than Chinese average. Mongolians aren't included for some reason, and neither are many Siberians such as Selkups, Nganassans and Yakuts, but they should all have considerable ANE, especially the Siberians.

South Asians like Sindhi are getting 2-3% SSA/Yoruba in this test which doesn't usually happen. There must be something in there that can't be covered well by the references in this test, also reflected by them getting over 30% ANE which seems an excessive number for South Asians. Sindhi had less Ma-1 affinity in the related study's f4 test than Evens who get only 20% here. Either the Han, Bedouin and Yoruba references are improper for South Asians and result in them getting too high ANE numbers, or some ANE in Evens and related populations is hidden by something, most likely Han.

[Graham]

Using proper ancient samples is the future and has barely started.

[ZephyrousMandaru]

There's a new experiment going on, goal seems to be to calculate "real" ANE, ENA, Middle Eastern and Sub-Saharan ancestry by fitting populations between Mal'ta, Han Chinese, Bedouins and Yoruba. I'll add a reminder that the accuracy of these results is based on the assumption that Bedouins and Han do not have any ANE. Results Polako posted so far:

Sample: ANE (MA-1), ENA (Han), ME (Bedouin), SSA (Yoruba)
Abhkasian: 0.195, 0.055, 0.75, 0
Abhkasian: 0.176, 0.055, 0.768, 0
Chechen: 0.249, 0.089, 0.662, 0
Chechen: 0.24, 0.094, 0.665, 0
Cyprian: 0.141, 0.026, 0.833, 0
Cyprian: 0.107, 0.048, 0.845, 0
Lezgin: 0.279, 0.066, 0.655, 0
Lezgin: 0.283, 0.055, 0.662, 0
Afghan_Pashtun: 0.288, 0.19, 0.521, 0

https://docs.google.com/spreadsheet/...&rm=full#gid=0

The Bedouin sample in the K13 test is actually 6% Northeast African (peaks in Gumuz and Ari cultivators) and 3% SSA (peaks in Yoruba), so one would need to be noticeably SSA-admixed for it to show in this kind of test.

The Bedouins and Han aren't ANE free per say, but they do have the least amount of ANE, unless we have ancient DNA from the Middle East or East Asia. Both score low values for ANE typically only around 1-3% ANE. Using any other population may underestimate or overestimate ANE admixture.

[Argang]

The Bedouins and Han aren't ANE free per say, but they do have the least amount of ANE, unless we have ancient DNA from the Middle East or East Asia. Both score low values for ANE typically only around 1-3% ANE. Using any other population may underestimate or overestimate ANE admixture.

Han score low amounts of ANE in this test because they are reference samples for "ENA", so they are of course going to show up as fully or almost fully as the component they form. In formal f3-tests Han are considerably more like MA-1 and thus the "ANE" component it alone represents than Bedouins, and it can't really be said they are less ANE-like than Makrani, who get 28% ANE in the k4-admixture test with Han as ENA-reference. Balochi should be on the level of East Asians in ANE-likeness too.


http://www.nature.com/nature/journal...13025_SF5.html

Having Han represent ENA quite clearly overestimates ANE for South Asians and/or underestimates ANE for East Asians. I would say evidence points towards something much more southern than Han being a better proxy for ENA, but I'm not confident that any current population would fully be.

edit. Polako added results for Ket and Selkup. He doubted their accuracy due to small European ancestry in these populations, but I don't think it matters much because in fact Kets are supposed to be more like Mal'ta than any Europeans or Central Asians according to the sample's introductionary study, and both Kets and Selkups are supposed to be more like MA-1 than any population in this K4-run except Karitiana. They turned out a bit more ANE than Kalash, which is OK in my book and maybe the difference could have been even bigger.

A "pure" modern siberian sample like Nganassans should be on the level of Adygei or Pathan, though in this admixture run the two have very different levels of ANE, most likely due to South Asians getting exaggarated numbers.

[Black Wolf]

Polako is now offering a service called Eurogenes Genetic Ancestry Detective and is charging $90 US for it and it seems quite good. I payed for it and I am now receiving some interesting results. He is not completely done my analysis yet but here is some of what he had said about my results when I asked him about them so far.

''You appear to have around 50% North Eurasian hunter-gatherer ancestry (that's WHG + ANE). So the rest is Neolithic farmer stuff, although that contains an unknown amount (probably over 10%) of WHG-like ancestry too. I can't estimate that now, but I'll try in the near future.

And yes, the Siberian segments are very real. They're ancient, and probably linked to the spread of Uralic languages into Northeastern Europe during the metal ages, but they are the reason you're sitting left of the red line.

In fact, if I took them out of your data (or masked them), this would shift you to the right across the line.''

http://bga101.blogspot.ca/2014/03/eu...detective.html