Showing posts with label cluster analysis. Show all posts
Showing posts with label cluster analysis. Show all posts

Thursday, 13 August 2015

Revisiting Kurdish autosomal DNA (Eurogenes version)

Updated: 30.12.2015

Today,  I want to present Kurdish results using Eurogenes K13, presented on GEDmatch.com. Thanks to all the contributors of this post. This is an update of the previous experiment with using a different calculator and more individuals included. The calculator is supposed to be without calculator effect, thus it should present more clear results.

Kurdish sub-groups
Êzidî: 1
Feyli: 4
Kurmanji: 22
Sorani: 11
Zazaki: 18

Neighbouring peoples
Armenian: 20
Assyrian: 11
samples

ADMIXTURE results using Eurogenes K13. (Population Averages)

There is not much difference between sub-groups other than expected gradual variation by the geography. Overall they're pretty close. Feylis score the highest West+South Asian additionally the lowest West Med+North Atlantic. As shown at previous experiments, It is really hard to distinguish between Kurmanjis and Zazakis from overall admixture makeup so I'll take them as one. While Zazakis are just a hair lower, Kurmanjis score the highest North Atlantic+West Med and the lowest Red Sea, while Zazakis score the lowest South Asian, again Kurmanjis follow Zazakis just a hair behind. Kurmanjis, Soranis, Zazakis score similar level of Baltic admixture.


Feylis= Southern
Sorani = Central, Eastern, Southern
Kurmanji = Northern, North-eastern, North-western, Western, Far-western, additionally Central Anatolia and Xorasan exclaves
Zazaki = Far-Western, North-western, North-eastern, Western


Distances between Kurdish groups and their neighbours:

Kurds_F Kurds_S Kurds_E Kurds_K Kurds_Z Iranians Armenians  Assyrians Turkish
Kurds_F 0 3.9 4.1 4.6 4.5 3.1 13.3 13.6 14.9
Kurds_S 3.7 0 3.3 2.4 2.4 4.0 11.2 12.0 11.6
Kurds_E 4.1 3.3 0 3.5 2.9 5.6 9.9 10.9 12.7
Kurds_K 4.9 2.4 3.5 0 1.3 5.8 10.7 12.2 10.8
Kurds_Z 4.4 2.4 2.9 1.3 0 5.6 10.2 11.4 11.3
Iranians 3.1 4.0 5.6 5.8 5.6 0 13.9 13.8 14.2
Armenians 13.4 11.2 9.9 10.7 10.2 13.9 0 4.2 11.4
Assyrians 13.8 12.0 10.9 12.2 11.4 13.8 4.2 0 13.2
Turkish 14.9 11.6 12.7 10.8 11.3 14.2 11.4 13.2 0

Overall the closest people to Kurdish groups are their linguistic cousins, Iranians. Then Armenians, Assyrians, Turks. Syrians, Iraqis, Georgians are quite distant from Kurds hence excluded.

Despite having different religious beliefs and divided by international borders, Kurds are genetically quite homogeneous. When comparing Kurdish groups with their neighbours, there is no any sub clusters appear because of aforementioned reason. They all fall in the same cluster except a few admixed individuals. Therefore, We need to exclude neighbours to get finer details.

There are only 3 sub-clusters when all samples are considered(n=56). #2 seems like individuals with recent Armenian/-like ancestry. Thanks to more samples, this time division between  clusters is finer and there is no clear cut borders like previous experiment. We can say first group is merely Northern or second group is merely Southern. Because both clusters have all Kurdish groups:

#1 Northern/Western
#2 Individuals with recent foreign admixture
#3 Southern/Eastern
(see Results by class: )


Distances between the class centroids:

1
2
3
1
0
7.1
4.4
2
7.1
0
7.0
3
4.4
7.0
0



Class centroids:
Class
1
2
3
North Atlantic
3.9
3.7
2.5
Baltic
3.9
2.0
5.7
West Med
7.4
9.2
7.0
West Asian
40.4
38.1
37.5
East Med
29.1
34.8
30.4
Red Sea
5.1
4.3
6.1
South Asian
7.4
5.7
8.2
East Asian
0.4
0.3
0.3
Siberian
1.0
0.5
0.9
Amerindian
0.7
0.6
0.6
Oceanian
0.2
0.2
0.2
Northeast African
0.4
0
0.3
Sub Saharan
0.1
0.6
0.3

Results by class:
Class
       1
2
3
KRD000K
 KRD002K
 KRD003K
KRD005K
KRD012Z
KRD006K
KRD007K
KRD019K
KRD038K
KRD033K
KRD044K
KRD051K
KRD063K
KRD060K
KRD091K
KRD093K
KRD100K
KRD098K
KRD158K
KRD116K
KRD184K
KRD166K
KRD045E
KRD185K
KRD008Z
KRD001Z
KRD058Z
KRD004Z
KRD059Z
KRD039Z
KRD064Z
KRD048Z
KRD070Z
KRD077Z
KRD072Z
KRD078Z
KRD076Z
KRD086Z
KRD101Z
KRD029S
KRD102Z
KRD032S
KRD104Z
KRD047S
KRD030S
KRD061S
KRD031S
KRD089S
KRD138S
KRD092S
KRD192S
KRD148S
KRD036F
KRD034F
Kurds_E
KRD035F
Kurds_K
KRD090F
Kurds_Z
Kurds_F
Kurds_S





Same dendrogram instead of ID codes, member's locations



Click image to expand

Partly admixed individuals:

Euclidean distance:
KRD002K 
KRD012Z
ARM13
ARM19
ASY08
ARM
6,2
6,0
4,9
6,8
5,1
ASY
8,9
6,7
7,3
8,3
4,6
Kurds_F
10,0
8,2
10,8
9,2
9,7
Kurds_S
8,1
7,0
8,5
7,4
8,1
Kurds_E
6,3
4,9
7,7
6,8
7,1
Kurds_K
6.9
7,0
7,4
6,7
8,3
Kurds_Z
6,7
6,1
7.0
6,0
7,5

Participant Map:

Friday, 19 September 2014

Participants distances

Revisiting Kurdish Autosomal DNA is updated with new 4 Zaza samples. Thought I should post participants distances to major Kurdish groups and individuals.

You can download the results here

Corduene.blogspot Participant Map

Sunday, 9 February 2014

Revisiting Kurdish autosomal DNA

Update: 19.09.2014

I thought I should merge all samples of KurdishDNA and this blog and try to shed some light on. Here's the total of 34 samples.

Ezidi: 2
Feyli: 3
Kurmanji: 15
Sorani: 10
Zazaki: 9
of samples

First all major subgroups' ADMIXTURE values from Dodecad K12b calculator.(Population Averages)


There is not much of difference between sub-groups other than expected gradual variation by the geography. Feylis score the highest Gedrosia+pseudo South Asian+lowest Mediterranean and Zazaki's highest Caucasus+lowest pseudo South Asian, Kurmanjis similar to Zazakis additionally have highest Atlantic-Mediterrean and North European components.



Distances between dialect and religious sub-groups

Kurds_F Kurds_S Kurds_E Kurds_K Kurds_Z
Kurds_F 0 2,809 2,243 5,920 6,047
Kurds_S 2,809 0 2,665 5,109 5,540
Kurds_E 2,243 2,665 0 5,788 6,250
Kurds_K 5,920 5,109 5,788 0 1,519
Kurds_Z 6,047 5,540 6,250 1,519 0

There appears to be only 3 sub-groups when all samples are considered(n=34),  #2 is a Kurdish exclave:
#1 Northern/Western (Kurmanji, Zazaki)
#2 Central Anatolian Kurds (Kurdên Anatoliya Navîn)
#3 Southern/Eastern (Ezidi, Sorani, Feyli)
(see Results by class: )

Reminder this North/South divide is more pronounced because of the sample bias. See my previous post about Participants Map, most northern samples are from Dersim and most southern samples are from Silêmanî. If there were more samples between those areas gradual difference would be much more finer.

Distances between the class centroids:

       1 2 3
1 0 4,825 5,350
2 4,825 0 6,730
3 5,350 6,730 0

Class centroids: Class 1 2 3
Caucasus 42,141 38,678 38,820
Gedrosia 24,215 22,863 27,667
Southwest Asian 13,323 14,265 14,865
Atlantic Med 7,912 9,588 6,842
North European 6,927 9,260 5,592
South Asian 2,423 2,500 2,953
Siberian 0,948 0,780 1,150
East Asian 0,733 1,080 0,532
Southeast Asian 0,316 0,130 0,355
Northwest African 0,776 0,425 0,777
East African 0,258 0,360 0,338
Sub Saharan 0,063 0,000 0,105

Results by class:
Class 1 2 3
KRD0000K KRD0003K KD008E
KRD0001Z KRD0010K KD043E
KRD0002K KRD0011K KD009S
KRD0004Z KD010S
KRD0005K KD012S
KRD0006K KD013S
KRD0007K KD014S
KRD0008Z KD032S
KRD0009K KD033S
KRD0012Z KD034S
KRD0019K KD048S
KRD0039Z KD026F
KRD0044K KD027F
KRD0048Z KD025F
KRD0051K Kurds_F
KRD0052K Kurds_S
KRD0058Z Kurds_E
KRD0059Z
KRD0064Z
KD007K
KD023K
KD011S
Kurds_K
Kurds_Z





Click image to expand