Running Gedmatch calculators on the command line with stevenliuyi/admix

[Leto]

Lukasz was very quick to add those pops compared to adding what I send but I have a few questions:

Why does the Even have so much European ancestry? Why is the Yukagir Forest almost half European? The same goes for the Aleut average. Something must be wrong with them.

[Lucas]

Lukasz was very quick to add those pops compared to adding what I send but I have a few questions:

Why does the Even have so much European ancestry? Why is the Yukagir Forest almost half European? The same goes for the Aleut average. Something must be wrong with them.

It is not fault of those scripts. In K36 those samples are identical in terms of Euro ancestry. They are mestizos simply. But some full Siberians are among them still.

[Leto]

It is not fault of those scripts. In K36 those samples are identical in terms of Euro ancestry. They are mestizos simply. But some full Siberians are among them still.

Well, I admit I don't know too much about those tiny ethnic groups. Let's not add Yukagir Forest though.

Add these please

Saami_Sweden,4.28,23.14,0.03,0.32,11.85,55.53,0.35,0.01,0.00,4.49,0.00,0.00
Khanty,8.11,47.13,0.00,0.14,0.92,33.67,1.38,0.01,0.04,8.57,0.01,0.02
Thai,3.49,1.38,0.07,57.25,0.68,0.73,12.36,0.15,0.45,22.17,1.18,0.10
Yukagir_Tundra,0.54,68.96,0.06,0.29,0.02,3.30,0.72,0.01,0.00,26.06,0.00,0.04

Saami_Kola is like 1500 people, we can dispense with them too.

[Komintasavalta]

is this script much faster than the regular DiyDodecad, or you have very high CPU/RAM?

could you run these Lithuanians trough k13? https://figshare.com/articles/datase...P_data/7964159

It takes about 3 seconds per sample when I use GNU Parallel to run 10 parallel jobs, and 6 seconds otherwise.

Results of all 412 samples: https://pastebin.com/raw/xDUUK2pA.

In the PCA below, the outliers in the top left corner of the plot got something like 99.5% or 99.7% of Baltic. I don't know if it's because they were used as reference samples in K13. LTG-356 is an outlier because it only got 21.48% North_Atlantic, 74.47% Baltic, and 4.05% West_Med, and 0% all other components. LTG-441 is closest to Russian_Kargopol and LTG-566 is closest to Russian_average.

But the Khanty one is broken, please fix it.

Sorry, I fixed it now.

[vbnetkhio]

It takes about 3 seconds per sample when I use GNU Parallel to run 10 parallel jobs, and 6 seconds otherwise.

Results of all 412 samples: https://pastebin.com/raw/xDUUK2pA.

In the PCA below, the outliers in the top left corner of the plot got something like 99.5% or 99.7% of Baltic. I don't know if it's because they were used as reference samples in K13. LTG-356 is an outlier because it only got 21.48% North_Atlantic, 74.47% Baltic, and 4.05% West_Med, and 0% all other components. LTG-441 is closest to Russian_Kargopol and LTG-566 is closest to Russian_average.





Sorry, I fixed it now.

wow, thanks so much, it would've taken me ages to do this.
these were published long after k13 was made, maybe these were reused from the older set of Lithuanians or it's their close relatives.

[Komintasavalta]

Why does the Even have so much European ancestry?

There's some mixed Even samples that I probably should've removed. But all Aleut and Forest Yukaghir samples have at least 20% Baltic:

	North_Atlantic	Baltic	West_Med	West_Asian	East_Med	Red_Sea	South_Asian	East_Asian	Siberian	Amerindian	Oceanian	Northeast_African	Sub-Saharan
Even:Nlk3	0.00	0.00	0.61	0.00	0.00	0.00	1.64	11.72	83.00	2.22	0.82	0.00	0.00
Even:Nlk5	4.61	20.83	1.61	0.59	1.61	3.26	0.00	11.25	53.06	2.20	0.17	0.80	0.00
Even:Nlk6	2.43	11.94	0.00	0.00	0.53	0.00	2.42	9.46	71.25	1.78	0.00	0.20	0.00
Even:Nlk10	8.12	22.87	6.11	4.91	5.18	1.62	0.66	5.92	43.48	0.43	0.00	0.00	0.70
Even:Nlk14	14.59	21.61	6.28	0.00	2.32	0.00	1.04	5.13	45.46	1.87	1.61	0.00	0.11
Even:Nlk16	3.69	27.20	3.00	2.70	4.69	1.90	0.26	7.11	47.45	1.35	0.66	0.00	0.00
Even:Nlk18	0.00	1.53	0.26	0.00	0.00	0.00	1.53	9.38	84.79	1.37	0.86	0.29	0.00
Even:Nlk19	17.26	35.05	5.01	4.75	1.30	1.03	2.43	5.21	25.75	0.35	1.76	0.06	0.06
Yukagir_Forest:Nel13	14.97	24.73	4.83	2.67	2.31	0.00	0.00	7.37	40.01	1.92	0.00	0.00	1.19
Yukagir_Forest:Nel15	11.83	25.15	2.21	2.86	3.88	0.00	1.34	4.80	42.28	2.70	1.27	0.94	0.75
Yukagir_Forest:Nel16	11.94	31.40	5.73	0.00	2.53	1.80	2.03	2.54	40.82	1.21	0.00	0.00	0.00
Yukagir_Forest:Nel17	12.52	26.98	2.92	0.00	1.91	1.26	1.20	3.03	45.13	3.36	0.43	1.01	0.24
Yukagir_Forest:Nel19	14.62	32.52	2.83	0.00	4.86	0.00	2.73	4.17	36.62	1.02	0.00	0.00	0.63
Aleut:Ale20	8.81	22.08	0.33	5.90	0.05	0.05	1.58	6.92	22.90	30.77	0.12	0.16	0.35
Aleut:Ale22	11.05	21.55	0.18	4.70	0.00	0.00	1.36	6.00	27.08	27.92	0.00	0.00	0.15
Aleut:Ale33	21.16	43.78	3.59	2.95	0.00	1.20	0.00	0.67	13.32	11.49	0.68	0.00	1.16
Aleut:Ale34	14.66	31.39	0.00	3.29	0.00	0.33	0.99	4.58	22.92	21.06	0.31	0.00	0.47
Aleut:Ale35	13.88	39.52	6.91	5.68	0.00	0.00	2.15	1.87	15.55	12.97	0.91	0.45	0.09
Aleut:Ale36	21.46	46.28	8.39	4.78	7.91	0.00	1.34	0.17	5.58	3.35	0.73	0.00	0.00
Aleut:Ale37	15.91	42.25	2.12	4.98	3.09	1.52	2.21	1.34	11.76	13.18	0.90	0.45	0.28

There were already a bunch of mixed Aleuts in 1909:

http://collection.kunstkamera.ru/en/entity/OBJECT/31429
http://collection.kunstkamera.ru/en/.../OBJECT/325966
http://collection.kunstkamera.ru/en/.../OBJECT/325963
http://collection.kunstkamera.ru/en/.../OBJECT/326282

Let's not add Yukagir Forest though. ...

Saami_Kola is like 1500 people, we can dispense with them too.

Yukaghiric people are interesting because they were the indigenous population of Yakutia before the Turkic expansion. Also Yukaghiric languages are possibly the closest relatives of Uralic languages, and in the Uralo-Siberian theory of Fortescue, Yukaghiric languages play an important role in connecting Uralic languages to Siberian and Eskimo langugages. One of the most cited papers by the admin of the Finnish anthroforum is titled "Early contacts between Uralic and Yukaghir". The Ymyyakhtakh culture which spread as far as Finland was possibly Yukaghiric, or if Bolshoy Oleniy Ostrov wasn't Uralic, then it might have also been Yukaghiric.

Saami is actually a meta-ethnos which is about as old as the Germanic or Finnic meta-ethnos, and that speaks about 10 languages with limited mutual intelligibility. There are even 5 different Saami languages that are native to the area of Murmansk Oblast. Saami are also genetically very diverse, but I currently only have samples of Kola Saami, Swedish Saami (which are probably Northern Saami), and Finnish Saami from an area that speaks Northern Saami. G25 and the Reich dataset and other sources are missing samples for Inari Saami, Skolt Saami, Southern Saami, Ume Saami, Pite Saami, and Lule Saami (unless the Finnish Saami samples are actually Inari Saami and not Northern Saami). In G25 and K13 updated, there's now samples for most European peoples that speak their own language, but they're still missing samples for most Saami languages. Also Saami are a unique people because they are one of the few arctic reindeer herding people of Europe, and they were possibly the last true hunter-gatherers of Europe (even though it might also have been Sikhirtya who were marine mammal hunters that lived in Nenetsia before the Nenets expansion, or it might have been Mansi who used to live in Komi Republic).

Below is a PCA of population averages from K13 updated along with my new samples. I included the 200 samples that were the closest to a Swedish Saami sample when accounting for FST. Without Saami_Kola, there would actually be a huge gap between Vepsians and Saami_SWE (especially because this is also missing Komis and Russian_Arkhangelsk_Leshukonsky):



This map by Nykyus shows Yukaghir tribes in green:


https://forum.paradoxplaza.com/forum...#post-25248469

[Leto]

Well, thanks for the valuable information, I was aware of several Saami subgroups but in Murmansk oblast they are all almost extinct by now, perhaps they were never numerous to begin with. In Finland, Sweden and Norway they are a bit better off.

The Yukagir avg that is currently in the spreadsheet should be deleted, it's from the original calculator-affected sheet.

[Komintasavalta]

Here's averages from Cardona et al. 2014 ("Genome-Wide Analysis of Cold Adaptation in Indigenous Siberian Populations") (https://www.ncbi.nlm.nih.gov/geo/que...i?acc=GSE73996):

#K13
Altai_Teleut,6.66,15.17,0.01,8.69,0.00,0.22,2.51,19.64,42.30,3.62,0.63,0.35,0.21
Komi_Siberian,15.73,44.89,2.09,4.75,0.37,0.21,1.90,0.73,26.20,2.52,0.30,0.11,0.19
Nenets_Forest,1.42,19.73,0.00,1.58,0.00,0.04,3.21,0.97,67.21,5.01,0.54,0.02,0.28
Nenets_Tundra,1.70,18.42,0.21,1.76,0.00,0.04,3.69,1.67,67.12,4.84,0.32,0.11,0.13

#K15
Altai_Teleut,6.41,1.97,3.05,15.08,0.00,5.36,0.00,0.10,2.71,19.81,41.05,3.52,0.54,0.20,0.21
Komi_Siberian,13.71,8.14,16.28,28.19,0.75,1.23,0.08,0.02,1.80,0.94,25.73,2.69,0.28,0.06,0.11
Nenets_Forest,2.28,0.19,1.11,21.88,0.00,0.03,0.00,0.00,2.46,0.99,65.88,4.52,0.46,0.03,0.15
Nenets_Tundra,3.15,1.60,2.00,20.69,0.12,0.27,0.00,0.03,2.90,1.63,62.83,4.38,0.28,0.05,0.06

#K12b
Altai_Teleut,8.46,33.04,0.12,2.22,3.20,17.97,1.45,0.02,0.09,30.88,2.32,0.24
Komi_Siberian,6.10,27.21,0.12,0.05,10.35,46.18,1.16,0.13,0.32,4.61,3.73,0.03
Nenets_Forest,4.22,65.36,0.00,0.03,0.05,19.28,0.83,0.01,0.00,10.18,0.02,0.03
Nenets_Tundra,3.73,61.84,0.03,0.17,1.16,20.90,0.83,0.03,0.07,10.57,0.60,0.07

The only samples which I omitted from the averages were the one Tundra Nenets sample and one Forest Nenets sample which plot the furthest right here:



I have made a library for shells where I have functions with 1-4 letter names for performing most common tasks. So I actually ran the calculators using a oneliner like this:

x=cold;p Teleut Forest_Nentsi Komi Tundra_Nentsi|rp g/f/karafetpop>$x.pick;keep2 g/p/karafet $x;rmm $x.23;zs 2 $x.fam|, "plink --bfile $x --keep <(awk '\$2==x' x={} $x.fam) --recode 23 --out $x.23/{}";for c in K13 K15 K12b;do rmm $x.$c;p $x.23/*.txt|, "admix -f {} -m $c|grep %>$x.$c/{/.}";for f in $x.$c/*;do rr \ <$f|rc %|jk|aak `be<<<$f`;done|a1k '{print a[$1]":"$0}' <(-skg/calc/$x.$c;1nk g/calc/$x.{i,c}|r ':[^,]*'|tav ,|nats>g/calc/$c.$x.a;done

[vbnetkhio]

It takes about 3 seconds per sample when I use GNU Parallel to run 10 parallel jobs, and 6 seconds otherwise.

Results of all 412 samples: https://pastebin.com/raw/xDUUK2pA.

In the PCA below, the outliers in the top left corner of the plot got something like 99.5% or 99.7% of Baltic. I don't know if it's because they were used as reference samples in K13. LTG-356 is an outlier because it only got 21.48% North_Atlantic, 74.47% Baltic, and 4.05% West_Med, and 0% all other components. LTG-441 is closest to Russian_Kargopol and LTG-566 is closest to Russian_average.





Sorry, I fixed it now.

I ran these which were missing in your sheet:

SZ/RA:LTG-1158,32.71,49.24,6.55,3.32,4.84,0.02,2.25,0.31,0,0.76,0,0,0
RA:LTG-429,30.14,52.01,7.78,2.25,2.82,0,0.57,0.12,2.05,1.31,0.25,0.16,0.54
PZ:LTG-648,28.11,53.41,9.26,2.05,4.3,0.01,1.47,0.06,0,0.53,0.18,0.38,0.23
RA:LTG-423,28.42,54.26,7.64,2.04,3.78,0,0.26,0.56,1.02,1.23,0.19,0.42,0.17
PZ:LTG-435,25.91,56.05,10.82,3.96,0.9,0,0.42,0,0.15,0.94,0.84,0,0
RA:LTG-632,27.22,56.52,6.63,3.29,1.12,0,1.91,0,1.98,0.56,0.4,0.02,0.35
PZ:LTG-436,26.28,57.36,6.76,6.18,0,0.03,0.14,0,1.36,0.48,0.55,0.86,0
PZ:LTG-645,29.68,57.36,7.14,1.53,0.24,0,1.42,0,0.49,0.3,0.53,0.01,1.3
PZ:LTG-655,25.91,59.48,5.69,4.59,0.04,0.55,0.72,0,1.11,1.72,0.04,0,0.16
PZ:LTG-524,26.09,59.52,6.68,3.79,0.01,0,1.3,0,1.41,0.18,0,0.97,0.03
PZ:LTG-631,25.07,61.58,6.77,1.93,1.8,0.01,0.28,0,0,1,0.68,0,0.88

you can check if i did the averages correctly in the other thread, just in case

[Leto]

The Nenets averages should be added to D K12b.