PDA

View Full Version : Kurd-Brahui Genetics with qpAdm & Dstats



Kurd
01-24-2016, 07:23 PM
I recently posted historical evidence pointing to Kurd ancestry for Brahui on other threads in this section, including some information on the Kurd Brahui tribe in Pakistan. Although we are not certain which population brought the Brahui language to Pakistan, the following stats show that present Brahui are genetically very Kurd like. Now I take a look at the Kurd-Brahui genetic connection using formal stats.

The following qpAdm runs indicated that Brahui can simply be modeled as a 2 way Iraqi Kurd - Sindhi or Pathan, with about 90% Iraqi Kurd and about 10% Sindhi or Pathan. What I found surprising was the fixed paths shows Kurd-Brahui as a near clade. In fact the fixed path showed that Brahui can also be modeled as 100% Iraqi Kurd with a reasonable chisq and tail prob. For those not familiar, the higher the tail prob, and the lower the chisq, the better the fit.



BRAHUI
%
Chisq
Tail Prob


Kurd_C2
87.70%
1.04
90.40%


Pathan
12.30%










BRAHUI
%
Chisq
Tail Prob


Kurd_C2
100.00%
1.44
92.00%








BRAHUI
%
Chisq
Tail Prob


Kurd_C2
87.90%
1.01
90.80%


Sindhi
12.10%

Kurd
01-24-2016, 07:24 PM
Here is the raw data in support of the above. I will try to post some Dstats later.

f4rank: 2 dof: 3 chisq: 0.963 tail: 0.810136551 dofdiff: 5 chisqdiff: -0.963 taildiff: 1
B:
scale 1.000 1.000
Atayal_Coriell -0.143 0.298
MbutiPygmy 2.216 0.183
Karitiana 0.169 0.408
Papuan 0.063 -1.751
Ulchi -0.196 1.283
A:
scale 1477.764 8606.036
.Kurd_C2 0.189 0.385
Pashtun_Afghan -0.986 1.409
Sindhi -1.411 -0.931


full rank 1
f4info:
f4rank: 3 dof: 0 chisq: 0.000 tail: 1 dofdiff: 3 chisqdiff: 0.963 taildiff: 0.810136551
B:
scale 1.000 1.000 1.000
Atayal_Coriell -0.145 0.254 0.807
MbutiPygmy 2.216 0.160 -0.119
Karitiana 0.161 0.226 2.061
Papuan 0.056 -1.777 0.235
Ulchi -0.195 1.304 -0.180
A:
scale 1476.501 8761.180 10778.030
.Kurd_C2 0.205 0.209 -1.707
Pashtun_Afghan -0.984 1.424 0.056
Sindhi -1.411 -0.963 -0.287


best coefficients: 0.913 -0.116 0.203
ssres:
-0.000127899 0.000036699 -0.000313086 -0.000000535 -0.000000042
-0.840683900 0.241221608 -2.057923775 -0.003517584 -0.000277574

Jackknife mean: 0.856252945 0.021307371 0.122439683
std. errors: 0.249 0.610 0.463

error covariance (* 1000000)
62096 -109917 47822
-109917 371995 -262078
47822 -262078 214256


fixed pat wt dof chisq tail prob
000 0 3 0.963 0 0.913 -0.116 0.203 infeasible
001 1 4 1.223 0.874248 0.886 0.114 -0.000
010 1 4 1.012 0.907974 0.879 0.000 0.121
100 1 4 10.685 0 -0.000 3.143 -2.143 infeasible
011 2 5 1.437 0.920179 1.000 -0.000 -0.000
101 2 5 33.686 2.74931e-06 0.000 1.000 -0.000
110 2 5 172.373 0 0.000 -0.000 1.000
best pat: 000 0 - -
best pat: 010 0.907974 chi(nested): 0.049 p-value for nested model: 0.825337
best pat: 011 0.920179 chi(nested): 0.425 p-value for nested model: 0.514245

Kurd
01-24-2016, 07:25 PM
f4rank: 2 dof: 3 chisq: 0.708 tail: 0.87134585 dofdiff: 5 chisqdiff: -0.708 taildiff: 1
B:
scale 1.000 1.000
Atayal_Coriell -0.100 -0.444
MbutiPygmy 2.192 -0.347
Karitiana -0.049 1.240
Papuan 0.292 1.316
Ulchi -0.315 -1.189
A:
scale 1488.500 12324.637
.Kurd_C2 0.228 -1.105
Pathan -1.377 0.530
Pashtun_Afghan -1.025 -1.224


full rank 1
f4info:
f4rank: 3 dof: 0 chisq: 0.000 tail: 1 dofdiff: 3 chisqdiff: 0.708 taildiff: 0.87134585
B:
scale 1.000 1.000 1.000
Atayal_Coriell -0.120 0.434 1.636
MbutiPygmy 2.187 -0.185 0.383
Karitiana -0.054 1.790 0.581
Papuan 0.298 0.983 -0.914
Ulchi -0.336 -0.779 1.002
A:
scale 1499.552 10204.665 21192.227
.Kurd_C2 0.217 -1.538 -0.767
Pathan -1.383 0.303 -0.997
Pashtun_Afghan -1.019 -0.738 1.190


best coefficients: 0.995 0.630 -0.626
ssres:
-0.000200816 -0.000049478 -0.000211956 0.000003948 -0.000029910
-1.508483260 -0.371669213 -1.592162766 0.029659322 -0.224673767

Jackknife mean: 0.867853303 -0.037165893 0.169312590
std. errors: 0.367 1.403 1.685

error covariance (* 1000000)
134449 368935 -503384
368935 1968553 -2337488
-503384 -2337488 2840872


fixed pat wt dof chisq tail prob
000 0 3 0.708 0 0.995 0.630 -0.626 infeasible
001 1 4 1.037 0.904081 0.877 0.123 -0.000
010 1 4 1.223 0.874248 0.886 0.000 0.114
100 1 4 3.679 0 0.000 -3.755 4.755 infeasible
011 2 5 1.437 0.920179 1.000 -0.000 -0.000
101 2 5 110.266 3.59987e-22 0.000 1.000 -0.000
110 2 5 33.686 2.74931e-06 0.000 -0.000 1.000
best pat: 000 0 - -
best pat: 001 0.904081 chi(nested): 0.329 p-value for nested model: 0.565981
best pat: 011 0.920179 chi(nested): 0.400 p-value for nested model: 0.527062

Kurd
01-24-2016, 08:50 PM
Here are some Dstats comparing X against Balochi. Iraqi Kurds C1 and C2 show longer shared drift paths with Brahui than Balochi does with Brahui. I believe the stat would become more significant if I had more that 100K shared SNPs. Georgian shows about the same length shared drift path as Balochi. Anything below Georgian shares less drift with Brahui than Balochi does, although positions 4 - 7 are not significant.

Although Baloch share more recent drift with Brahui (past 1000 years), Dstats disregard post Kurd-Brahui, or post X-Brahui split accumulated mutations (drift), and thus consider the longer shared drift path leading up to the split



D ( Balochi, X, Brahui, Gorilla)


X
D
Z
SNPs


.Kurd_C2
-0.0025
-0.657
100621


.Kurd_C3
-0.0014
-0.357
100719


Georgian
0
0.003
152010


Abkhasian
0.0002
0.097
152010


Adygei
0.0007
0.457
152010


Kalash
0.0018
0.983
152010


Armenian
0.0018
1.048
152010


.Kurd_SE
0.0026
0.705
100610


Kurd_N
0.0037
1.931
114164


.Kurd_C1
0.0044
1.13
100542


Pathan
0.0044
2.783
152010


Azeri
0.006
3.291
71806


Pashtun_Afghan
0.0061
2.564
114103


Sindhi
0.0071
5.136
152010


Iranian
0.0072
4.228
152010


Makrani
0.0082
5.53
152010


Makrani
0.0082
5.53
152010


Turkish
0.0083
4.051
152010


Druze
0.0117
7.903
152010


Burusho
0.0119
8.214
152010


Brahmin
0.0123
6.107
83747


Tajik_Afghan
0.0146
7.17
114103


GujaratiD
0.0154
8.187
152010


Kallar
0.0201
7.718
83747


Bhil
0.0226
10.771
83747


Adi-Dravider
0.0242
9.23
83747


Saudi
0.0273
14.342
152010


Chenchu
0.0281
9.351
83747


Syrian
0.0284
15.276
152010


Irula
0.0293
10.59
83747


Birhor
0.0403
12.699
83747

Kaido
01-24-2016, 09:27 PM
Apologies if this has been mentioned on the forum elsewhere, but how do I read these results? What are the D and Z values?

Kurd
01-24-2016, 11:30 PM
Apologies if this has been mentioned on the forum elsewhere, but how do I read these results? What are the D and Z values?

Kaido,

This is directly from the Dstat Readme file:

DOCUMENTATION OF D-statistics (qpDstat):

The 4-population test, implemented here as D-statistics, is also a formal test for admixture based on a four taxon 4 statistic, which can provide some information about the direction of gene flow.
For any 4 populations (W, X, Y, Z), qpDstat computes the D-statistics as -
num = (w − x)(y − z )
den = (w + x − 2wx)(y + z − 2yz )

D = num/ den

The output of qpDstat is informative about the direction of gene flow. So for 4 populations (W, X, Y, Z) as follows -
If the Z-score is +ve, then the gene flow occured either between W and Y or X and Z
If the Z-score is -ve, then the gene flow occured either between W and Z or X and Y.

The ADMIXTOOLS package implements 5 methods described in Patterson et al. (2012) Ancient Admixture in Human History. Details of the methods and algorithm can be found in this paper.

Kurd
01-24-2016, 11:36 PM
These compare Pathans against various samples. The negative ones share more drift with Pathan than Pashtun_Afghan shares with Pathan, indicating that Afghan Pashtuns and Pakistani Pashtuns have slightly different drift histories. It looks like Andronovo and some Iraqi Kurd tribes have some deep connection to Pathan ancestry.



L1
L2
R1
OUT
D
Z
SNPs


Pashtun_Afghan
Andronovo
Pathan
Primate_Gorilla
-0.0059
-1.501
113527


Pashtun_Afghan
Sintashta_MBA
Pathan
Primate_Gorilla
-0.0056
-1.164
97964


Pashtun_Afghan
.Kurd_C3
Pathan
Primate_Gorilla
-0.0026
-0.597
97798


Pashtun_Afghan
.Kurd_C2
Pathan
Primate_Gorilla
-0.0024
-0.544
97708


Pashtun_Afghan
Kalash
Pathan
Primate_Gorilla
-0.0016
-0.638
114103


Pashtun_Afghan
Abkhasian
Pathan
Primate_Gorilla
-0.001
-0.395
114103


Pashtun_Afghan
Adygei
Pathan
Primate_Gorilla
-0.0009
-0.401
114103


Pashtun_Afghan
.Kurd_SE
Pathan
Primate_Gorilla
0.0001
0.018
97694


Pashtun_Afghan
Georgian
Pathan
Primate_Gorilla
0.0006
0.243
114103


Pashtun_Afghan
GujaratiB
Pathan
Primate_Gorilla
0.0011
0.419
114103


Pashtun_Afghan
GujaratiA
Pathan
Primate_Gorilla
0.0021
0.798
114103


Pashtun_Afghan
Armenian
Pathan
Primate_Gorilla
0.0032
1.29
114103


Pashtun_Afghan
Sindhi
Pathan
Primate_Gorilla
0.0032
1.391
114103


Pashtun_Afghan
.Kurd_C1
Pathan
Primate_Gorilla
0.0035
0.781
97628


Pashtun_Afghan
Burusho
Pathan
Primate_Gorilla
0.0042
1.907
114103


Pashtun_Afghan
Tajik
Pathan
Primate_Gorilla
0.0045
1.938
113999


Pashtun_Afghan
Balochi
Pathan
Primate_Gorilla
0.0047
1.948
114103


Pashtun_Afghan
Turkish
Pathan
Primate_Gorilla
0.0057
2.006
114103


Pashtun_Afghan
GujaratiC
Pathan
Primate_Gorilla
0.0058
2.216
114103


Pashtun_Afghan
Brahui
Pathan
Primate_Gorilla
0.0065
2.856
114103


Pashtun_Afghan
GujaratiD
Pathan
Primate_Gorilla
0.0066
2.478
114103


Pashtun_Afghan
Iranian
Pathan
Primate_Gorilla
0.008
3.195
114103


Pashtun_Afghan
Kotias
Pathan
Primate_Gorilla
0.0087
1.976
96552


Pashtun_Afghan
Tajik_Afghan
Pathan
Primate_Gorilla
0.0097
3.67
114103


Pashtun_Afghan
Adi-Dravider
Pathan
Primate_Gorilla
0.0104
2.569
47208


Pashtun_Afghan
Bhil
Pathan
Primate_Gorilla
0.0108
3.081
47208


Pashtun_Afghan
Druze
Pathan
Primate_Gorilla
0.0122
5
114103


Pashtun_Afghan
Chenchu
Pathan
Primate_Gorilla
0.0126
2.877
47208


Pashtun_Afghan
Makrani
Pathan
Primate_Gorilla
0.0135
5.596
114103


Pashtun_Afghan
Uzbek
Pathan
Primate_Gorilla
0.0148
5.924
114103


Pashtun_Afghan
Hazara
Pathan
Primate_Gorilla
0.0169
6.759
114103


Pashtun_Afghan
Paniyas
Pathan
Primate_Gorilla
0.0245
5.397
47208


Pashtun_Afghan
Sherpa
Pathan
Primate_Gorilla
0.0274
6.141
47208


Pashtun_Afghan
Syrian
Pathan
Primate_Gorilla
0.0299
11.066
114103


Pashtun_Afghan
BedouinB
Pathan
Primate_Gorilla
0.0344
12.648
114103


Pashtun_Afghan
Onge
Pathan
Primate_Gorilla
0.0414
8.404
47208

Kurd
01-25-2016, 04:34 AM
Here is a good qpAdm fit for Pathan



PATHAN
%
Chisq
Tail Prob


Kurd_C2
27.80%
0.285
96.30%


Andronovo
49.10%




Paniya
23.20%





codimension 1
f4info:
f4rank: 2 dof: 3 chisq: 0.285 tail: 0.962817085 dofdiff: 5 chisqdiff: -0.285 taildiff: 1
B:
scale 1.000 1.000
Atayal_Coriell 0.046 0.293
MbutiPygmy 1.767 -1.353
Karitiana 1.320 1.512
Papuan -0.286 -0.700
Ulchi 0.226 0.555
A:
scale 621.339 2139.511
.Kurd_C2 0.434 -1.478
Andronovo.SG 0.509 0.895
Paniyas -1.598 -0.124


full rank 1
f4info:
f4rank: 3 dof: 0 chisq: 0.000 tail: 1 dofdiff: 3 chisqdiff: 0.285 taildiff: 0.962817085
B:
scale 1.000 1.000 1.000
Atayal_Coriell 0.045 0.290 0.264
MbutiPygmy 1.770 -1.333 -0.250
Karitiana 1.314 1.533 0.899
Papuan -0.294 -0.701 1.624
Ulchi 0.229 0.546 -1.192
A:
scale 620.907 2151.652 23055.915
.Kurd_C2 0.431 -1.498 -0.756
Andronovo.SG 0.502 0.860 -1.418
Paniyas -1.601 -0.134 -0.647


best coefficients: 0.278 0.491 0.232
ssres:
-0.000015868 0.000011269 -0.000070630 -0.000063582 0.000046205
-0.330224342 0.234508478 -1.469868269 -1.323193206 0.961563659

Jackknife mean: 0.284367553 0.482354614 0.233277833
std. errors: 0.122 0.113 0.045

error covariance (* 1000000)
14919 -12837 -2081
-12837 12785 53
-2081 53 2029


fixed pat wt dof chisq tail prob
000 0 3 0.285 0.962817 0.278 0.491 0.232
001 1 4 15.157 0 3.715 -2.715 -0.000 infeasible
010 1 4 10.776 0.0291981 0.818 0.000 0.182
100 1 4 3.638 0.457273 0.000 0.743 0.257
011 2 5 17.286 0.00398894 1.000 0.000 -0.000
101 2 5 18.406 0.00247814 0.000 1.000 -0.000
110 2 5 253.356 0 0.000 -0.000 1.000
best pat: 000 0.962817 - -
best pat: 100 0.457273 chi(nested): 3.353 p-value for nested model: 0.0671024
best pat: 011 0.00398894 chi(nested): 13.648 p-value for nested model: 0.000220479

Kurd
01-25-2016, 12:28 PM
Pathan fits with Andronovo and Paniya fixed, but varying the 3rd population between Kurd, Georgian, Iranian, and Kotias CHG. The best fit was using Kurd.



PATHAN
%
Chisq
Tail Prob


Kurd_C2
27.80%
0.285
96.30%


Andronovo
49.10%




Paniya
23.20%










PATHAN
%
Chisq
Tail Prob


Georgian
24.20%
1.772
62.10%


Andronovo
49.20%




Paniya
26.60%










PATHAN
%
Chisq
Tail Prob


Iranian
23.90%
1.897
59.40%


Andronovo
51.00%




Paniya
25.00%










PATHAN
%
Chisq
Tail Prob


Kotias CHG
20.80%
3.754
28.90%


Andronovo
52..9%




Paniya
26.40%

Karim Dad
06-06-2016, 09:02 PM
Hi bro! I am Baloch from Balochistan i found your notes interesting, wanted to message you but as i have not posted anything here so i cant message you. Kindly if you send me your email so that i can ask you something about Kurd and Baloch genetics.
Regards
KarimDad

dp
06-06-2016, 09:28 PM
Kurd's email, based on his GedrosiaDNA admix calcuators on GEDmatch, is: [email protected]
dp

Hi bro! I am Baloch from Balochistan i found your notes interesting, wanted to message you but as i have not posted anything here so i cant message you. Kindly if you send me your email so that i can ask you something about Kurd and Baloch genetics.
Regards
KarimDad

Kurd
06-06-2016, 10:11 PM
Hi bro! I am Baloch from Balochistan i found your notes interesting, wanted to message you but as i have not posted anything here so i cant message you. Kindly if you send me your email so that i can ask you something about Kurd and Baloch genetics.
Regards
KarimDad

Salaam and welcome to the forum. You can email me at [email protected]

Karim Dad
06-06-2016, 10:21 PM
spas dasti xosh

Karim Dad
06-06-2016, 10:29 PM
check your mail, Kurd

J Man
06-10-2016, 04:21 AM
check your mail, Kurd

Have you had your Y-DNA tested?