Page 1 of 6 123 ... LastLast
Results 1 to 10 of 57

Thread: Pashtun Deep Ancestry Using qpAdm (5 Population Sources)

  1. #1
    Registered Users
    Posts
    4,115
    Sex

    Lightbulb Pashtun Deep Ancestry Using qpAdm (5 Population Sources)

    I will be posting qpAdm results for Meli and Surbakhun by tonight. What I have done is present 5 base populations to the program (think of them as base ingredients), and let the program pick the proportions of base ingredients that would make the best fits for Meli and Surbakhun.

    The base pops (ingredients) I have used are:

    1- Caucausus based ( a Kurd sample)
    2- Steppe based (an Andronovo sample)
    3- Pulliyar ( to represent the S Indian part of the admixture). Dravidian pops such as Pulliyar may have inhabited SC Asia in the past.
    4- Scythian IA ( as a possible vector for E Asian admixture in Pashtuns since Saka were local to C/SC Asia)
    5- Mongolian (as an alternative vector to Scythian for the E Asian admixture in Pashtuns, based on Mongolian invasions)

    I may offer Iranian as a 6th ingredient (vector for ME admixture in Pashtuns alternative to Kurds)

    The program will do the number crunching and present various proportions with their respective probabilities with I will post.

    The problem that I have with the modeling of Pashtuns that many have based on just 3 populations, Dai S Indian, and Andronovo is that it is a theoretical construct, as Dai are not a neighboring pop to Pashtuns, and there is no evidence of Dai interacting with Pashtuns.

    So why does this create a problem?
    Well, in addition to the problem with Dai described above, it forces the program to increase the Andronovo % to meet the demand of CHG based ancestry in Pashtuns, and neglects the fact that Pashtuns received a good portion of their CHG from W Asian based pops.

    So what happens when we add a W Asian pop, such as Kurd to the program to reflect the reality of Pashtuns receiving some of their DNA from W Asian pops?
    The program no longer needs to inflate the ANDRONOVO % to meet the CHG admixture requirements in Pashtun, as the program likes the W Asian pop for the CHG admixture in Pashtuns, which results in the ANDRONOVO % dropping. In this case the program picks either SCYTHIAN IA (for EHG and E Asian in Pashtuns above and beyond that present in the W Asian pop) , OR picks ANDRONOVO (for EHG in Pashtuns over and above that present in W Asian) AND MONGOLIAN (for E Asian in Pashtuns above and beyond that present in W Asian)


    By making available 5 or 6 populations to the program based on geography and Pashtun admixture allows the program to output combinations based on fits and assign them probabilities.

    I will post all the combinations and their fits / probabilities
    Last edited by Kurd; 05-13-2016 at 05:28 PM.

  2. The Following 10 Users Say Thank You to Kurd For This Useful Post:

     DMXX (05-13-2016), dp (05-14-2016), Hanna (05-13-2016), Kaido (05-13-2016), kenji.aryan (05-13-2016), khanabadoshi (05-13-2016), Mellifluous (05-13-2016), NK19191 (05-13-2016), Rukha (05-13-2016), surbakhunWeesste (05-13-2016)

  3. #2
    Gold Member Class
    Posts
    848
    Sex
    Location
    event horizon
    Ethnicity
    Ibex retter; Der Nomad
    Nationality
    Kochyan
    Y-DNA
    R1a......(Dad)
    mtDNA
    U2b2

    Afghanistan United States of America
    Quote Originally Posted by Kurd View Post
    I will be posting qpAdm results for Meli and Surbakhun by tonight. What I have done is present 5 base populations to the program (think of them as base ingredients), and let the program pick the proportions of base ingredients that would make the best fits for Meli and Surbakhun.

    The base pops (ingredients) I have used are:

    1- Caucausus based ( a Kurd sample)
    2- Steppe based (an Andronovo sample)
    3- Pulliyar ( to represent the S Indian part of the admixture). Dravidian pops such as Pulliyar may have inhabited SC Asia in the past.
    4- Scythian IA ( as a possible vector for E Asian admixture in Pashtuns since Saka were local to C/SC Asia)
    5- Mongolian (as an alternative vector to Scythian for the E Asian admixture in Pashtuns, based on Mongolian invasions)

    I may offer Iranian as a 6th ingredient (vector for ME admixture in Pashtuns alternative to Kurds)

    The program will do the number crunching and present various proportions with their respective probabilities with I will post.

    The problem that I have with the modeling of Pashtuns that many have based on just 3 populations, Dai S Indian, and Andronovo is that it is a theoretical construct, as Dai are not a neighboring pop to Pashtuns, and there is no evidence of Dai interacting with Pashtuns.

    So why does this create a problem?
    Well, in addition to the problem with Dai described above, it forces the program to increase the Andronovo % to meet the demand of CHG based ancestry in Pashtuns, and neglects the fact that Pashtuns received a good portion of their CHG from W Asian based pops.


    So what happens when we add a W Asian pop, such as Kurd to the program to reflect the reality of Pashtuns receiving some of their DNA from W Asian pops?
    The program no longer needs to inflate the ANDRONOVO % to meet the CHG admixture requirements in Pashtun, as the program likes the W Asian pop for the CHG admixture in Pashtuns, which results in the ANDRONOVO % dropping. In this case the program picks either SCYTHIAN IA (for EHG and E Asian in Pashtuns above and beyond that present in the W Asian pop) , OR picks ANDRONOVO (for EHG in Pashtuns over and above that present in W Asian) AND MONGOLIAN (for E Asian in Pashtuns above and beyond that present in W Asian)


    By making available 5 or 6 populations to the program based on geography and Pashtun admixture allows the program to output combinations based on fits and assign them probabilities.

    I will post all the combinations and their fits / probabilities
    Is this the reason why the Pathan sample showed ~60% Andronovo and some people were stuck on it. LOL

  4. The Following 2 Users Say Thank You to surbakhunWeesste For This Useful Post:

     DMXX (05-13-2016), khanabadoshi (05-13-2016)

  5. #3
    Registered Users
    Posts
    4,115
    Sex

    Quote Originally Posted by surbakhunWeesste View Post
    Is this the reason why the Pathan sample showed ~60% Andronovo and some people were stuck on it. LOL
    I believe so. Without the availability of a W Asian population in the run, which is unrealistic since I believe there were W Asian based populations in SC Asia in the past, the program is forced to increase Andronovo % to account for the caucausus based ancestry in Pashtuns. For a more realistic model, all populations which are possible contributors to Pashtun DNA based on geography and other known information should be made available in the run, so that the program can choose the proportions for the best fit.

    If the goal of the run is merely to determine percentages of basal/ancestral populations in Pashtun, then it is ok to use an ASI reference (Onge or a S Indian tribal), an EHG reference (Karelia HG), a CHG reference (Kotias), and an E Asian reference (Han or Dai or some other), but the mix will not reflect realistic interacitons

    EDIT: I just started a new thread for the qpAdm results under the Pashto section

  6. The Following 3 Users Say Thank You to Kurd For This Useful Post:

     DMXX (05-13-2016), khanabadoshi (05-13-2016), surbakhunWeesste (05-13-2016)

  7. #4
    Administrator
    Posts
    2,428
    Sex
    Y-DNA
    R2a*-M124 (L295-)
    mtDNA
    D4j5*

    England
    The "Scythian_IA" sample comes from the Volga region; odds are it will not resemble the Iron Age Saka found in places like Tajikistan. Going by the GD's in several calculators (punt K12 Ancient below), they look like some sort of mixture between the older Yamnaya-Afanasievo era and more recent Sintashta-Andronovo era steppe.

     

    # Population (source) Distance
    1 Scythian_IA_I0247 0.86
    2 Yamnaya_Samara_I0443 12.52
    3 Andronovo_SG_RISE505 13.05
    4 Poltavka_I0440 13.81
    5 Yamnaya_Samara_I0231 13.87
    6 Corded_Ware_Germany_I0104 14.17
    7 Yamnaya_Kalmykia_SG_RISE552 14.67
    8 Afanasievo_SG_RISE511 14.83
    9 Srubnaya_I0430 15.87
    10 Sintashta_MBA_RISE_386 15.99


    The physical anthropology indicates the Saka of South-Central Asia were quite different from the populations that resided on the steppes during the Bronze and Iron Age. Craniometry satisfactorily predicted Afanasievo is derived from Yamnaya years before 2015, so this snippet of evidence can't be ruled out so easily.

    If they're being used as a population source in qpAdm for Pashtuns, they'll demonstrate an exaggerated score due to their Yamnaya-type admixture (older steppe samples look more CHG-rich, EHG apparently functions as well as Mal'ta for replicating ANE admixture). In retrospect, this would explain why nearly everyone from Kurdistan to the Subcontinent has a smaller GD with Scythian_IA than the actual steppe source of proto-Indo-Iranian in Asia (Sintashta) or its' direct derivatives (Andronovo).

    Iranians and Kurds contain some steppe admixture themselves, so using them as a surrogate for more recent West Asian ancestry in Pashtuns would, in turn, cause a depreciation in their actual steppe ancestry. One workaround were if you used ADMIXTURE to identify those Iranians and Kurds with the least steppe admixture and utilise those?

    Another option might be to consider using Kotias, Sintashta (or the least East Eurasian Andronovo samples if deamination is an issue), Iraqi/Iranian Jew, Pulliyar and Korean? Depends what the aspirations are.

    [Edit]: Just read your post above, that second option probably doesn't serve the goal in that case.
    Last edited by DMXX; 05-13-2016 at 09:21 PM. Reason: Update from Kurd's post

  8. The Following 5 Users Say Thank You to DMXX For This Useful Post:

     Agamemnon (05-14-2016), khanabadoshi (05-13-2016), Kurd (05-13-2016), NK19191 (05-13-2016), surbakhunWeesste (05-13-2016)

  9. #5
    Registered Users
    Posts
    4,115
    Sex

    I let the program pick the proportions for the best fits. The results are sorted from best fit/ highest probability to worst fit/ lowest probability.

    With qpAdm, the higher the tail prob and the lower the chisq, the better the fit, and the more likely that the scenario is real. Although, I used Kurd, other caucausus based groups are a possibility also. So in this case one has to look at other circumstantial evidence such as language and other forms of evidence. I may post another set substituting Iranian for Kurd.


    First, Meli's fits.

    NO SAMPLE .Kurd_C2 Andronovo2 Palliyar Scythian_IA Mongolian CHISQ TAIL PROBABILITY
    1 MELI 45% 0% 33% 23% 0% 0.094 99.3%
    2 MELI 42% 5% 32% 21% 0% 0.071 96.5%
    3 MELI 67% 0% 28% 0% 5% 0.625 89.1%
    4 MELI 52% 16% 28% 0% 4% 0.348 84.0%
    5 MELI 63% 0% 37% 0% 0% 2.042 72.8%
    6 MELI 41% 23% 36% 0% 0% 1.592 66.1%
    7 MELI 0% 65% 35% 0% 0% 2.818 58.9%
    8 MELI 0% 49% 33% 19% 0% 2.091 55.4%
    9 MELI 0% 67% 30% 0% 2% 2.486 47.8%
    10 MELI 52% 16% 28% 0% 4% 2.028 36.3%
    11 MELI 90% 0% 0% 0% 10% 4.977 29.0%
    12 MELI 0% 92% 0% 0% 8% 6.26 18.1%
    13 MELI 0% 0% 34% 66% 0% 7.055 13.3%
    14 MELI 0% 100% 0% 0% 0% 10.715 5.7%
    15 MELI 50% 0% 0% 51% 0% 9.338 5.3%
    16 MELI 0% 61% 0% 39% 0% 10.17 3.8%
    17 MELI 0% 0% 0% 100% 0% 12.574 2.8%
    18 MELI 100% 0% 0% 0% 0% 14.014 1.6%
    19 MELI 0% 0% 100% 0% 0% 51.252 0.0%
    20 MELI 0% 0% 0% 0% 100% 925.579 0.0%

  10. The Following 4 Users Say Thank You to Kurd For This Useful Post:

     Kaido (05-13-2016), khanabadoshi (05-13-2016), Mellifluous (05-14-2016), surbakhunWeesste (05-13-2016)

  11. #6
    Registered Users
    Posts
    4,115
    Sex

    In the interest of transparency, here is the raw data for Meli

    0 .Melif 1
    1 .Kurd_C2 1
    2 Andronovo2 1
    3 Palliyar 5
    4 Scythian_IA 1
    5 Mongola 6
    6 MbutiPygmy 10
    7 Karitiana 12
    8 Onge 9
    9 Han 33
    10 Yoruba 70
    11 Ami 10
    jackknife block size: 0.05
    snps: 532428 indivs: 159
    number of blocks for block jackknife: 711
    dof (jackknife): 617.196
    numsnps used: 26214
    codimension 1
    f4info:
    f4rank: 4 dof: 1 chisq: 0.068 tail: 0.794603551 dofdiff: 3 chisqdiff: -0.068
    B:
    scale 1 1 1 1
    Karitiana 0.851 -1.227 1.645 -0.013
    Onge 0.573 1.841 1.1 0.204
    Han 1.435 0.01 -0.69 -1.154
    Yoruba -0.008 -0.325 0.008 1.496
    Ami 1.375 0.003 -0.781 1.179
    A:
    scale 160.759 1090.496 3009.107 9496.274
    .Kurd_C2 -0.35 -0.625 -1.185 -0.719
    Andronovo2 -0.148 -0.816 1.549 -1.481
    Palliyar 0.321 1.628 0.769 0.016
    Scythian_IA 0.218 -1.097 0.71 1.424
    Mongola 2.169 -0.299 -0.319 -0.512


    full rank 1
    f4info:
    f4rank: 5 dof: 0 chisq: 0 tail: 1 dofdiff: 1 chisqdiff: 0.068
    B:
    scale 1 1 1 1 1
    Karitiana 0.85 -1.233 1.638 -0.081 -0.259
    Onge 0.573 1.833 1.115 0.214 0.147
    Han 1.437 -0.01 -0.64 -1.166 1.08
    Yoruba -0.004 -0.347 0.073 1.477 1.641
    Ami 1.372 0.008 -0.81 1.186 -1.027
    A:
    scale 160.29 1097.704 2956.538 9757.235 32811.239
    .Kurd_C2 -0.34 -0.609 -1.096 -0.53 1.741
    Andronovo2 -0.139 -0.798 1.576 -1.335 0.279
    Palliyar 0.328 1.653 0.812 0.189 1.211
    Scythian_IA 0.225 -1.086 0.763 1.663 0.65
    Mongola 2.17 -0.284 -0.272 -0.37 0.04


    best coefficients: 0.403 0.036 0.331 0.235 -0.005
    ssres:
    0.000047014 0.0000926 0.000077789 0.000088494 0.000036777
    0.651714242 1.283638008 1.07832346 1.226724758 0.509810646

    Jackknife mean: 0.402419837 0.054087233 0.333538746 0.210183093 -0.000228908
    std. errors: 0.352 0.367 0.147 0.445 0.095

    error covariance (* 1000000)
    124006 -23451 -28667 -93411 21523
    -23451 134926 -23495 -106974 18994
    -28667 -23495 21646 41282 -10765
    -93411 -106974 41282 197785 -38682
    21523 18994 -10765 -38682 8931


    fixed pat wt dof chisq tail prob
    0 0 1 0.068 0 0.403 0.036 0.331 0.235 -0.005 infeasible
    1 1 2 0.071 0.965181 0.416 0.046 0.324 0.213 0
    10 1 2 0.348 0.840469 0.517 0.163 0.279 0 0.041
    100 1 2 2.028 0.362835 0.517 0.163 0.279 0 0.041
    1000 1 2 0.077 0 0.41 0 0.337 0.263 -0.01 infeasible
    10000 1 2 1.103 0 0 -0.006 0.439 0.663 -0.097 infeasible
    11 2 3 1.592 0.661265 0.413 0.227 0.359 0 0
    101 2 3 5.492 0 5.522 -6.446 0 1.924 0 infeasible
    110 2 3 4.928 0 1.043 -0.146 0 0 0.103 infeasible
    1001 2 3 0.094 0.992599 0.45 0 0.325 0.225 0
    1010 2 3 0.625 0.890627 0.674 0 0.282 0 0.045
    1100 2 3 3.221 0 1.445 0 0 -0.648 0.203 infeasible
    10001 2 3 2.091 0.553747 0 0.486 0.326 0.189 0
    10010 2 3 2.486 0.477757 0 0.673 0.303 0 0.024
    10100 2 3 2.814 0 0 2.307 0 -1.619 0.312 infeasible
    11000 2 3 1.103 0 0 0 0.438 0.658 -0.096 infeasible
    111 3 4 6.21 0 -3.979 4.979 0 0 0 infeasible
    1011 3 4 2.042 0.727977 0.63 0 0.37 0 0
    1101 3 4 9.338 0.0531932 0.495 0 0 0.505 0
    1110 3 4 4.977 0.28965 0.899 0 0 0 0.101
    10011 3 4 2.818 0.588723 0 0.649 0.351 0 0
    10101 3 4 10.17 0.0376599 0 0.612 0 0.388 0
    10110 3 4 6.26 0.18053 0 0.916 0 0 0.084
    11001 3 4 7.055 0.133034 0 0 0.341 0.659 0
    11010 3 4 46.592 0 0 0 1.074 0 -0.074 infeasible
    11100 3 4 9.667 0 0 0 0 1.084 -0.084 infeasible
    1111 4 5 14.014 0.0155181 1 0 0 0 0
    10111 4 5 10.715 0.0573348 0 1 0 0 0
    11011 4 5 51.252 7.68E-10 0 0 1 0 0
    11101 4 5 12.574 0.0277112 0 0 0 1 0
    11110 4 5 925.579 0 0 0 0 0 1
    best pat: 0 0 - -
    best pat: 1 0.965181 chi(nested): 0.003 p-value for nested model: 0.955574
    best pat: 1001 0.992599 chi(nested): 0.023 p-value for nested model: 0.880286
    best pat: 1011 0.727977 chi(nested): 1.949 p-value for nested model: 0.162722
    best pat: 10111 0.0573348 chi(nested): 8.673 p-value for nested model: 0.0032303

  12. The Following 3 Users Say Thank You to Kurd For This Useful Post:

     dp (05-14-2016), khanabadoshi (05-13-2016), surbakhunWeesste (05-13-2016)

  13. #7
    Registered Users
    Posts
    4,115
    Sex

    Kandhari's results are somewhat different from Meli, reflecting greater Caucausus based ancestry.

    NO SAMPLE .Kurd_C3 Andronovo2 Palliyar Scythian_IA Mongolian CHISQ TAIL PROBABILITY
    1 KANDHARI 67% 26% 0% 0% 7% 0.235 97%
    2 KANDHARI 91% 0% 0% 0% 9% 1.077 90%
    3 KANDHARI 54% 5% 8% 33% 0% 0.219 90%
    4 KANDHARI 57% 0% 8% 35% 0% 0.657 88%
    5 KANDHARI 58% 0% 0% 43% 0% 1.291 86%
    6 KANDHARI 76% 0% 0% 19% 5% 0.802 85%
    7 KANDHARI 88% 0% 4% 0% 8% 0.973 81%
    8 KANDHARI 67% 0% 6% 24% 3% 0.53 77%
    9 KANDHARI 52% 9% 0% 39% 0% 1.209 75%
    10 KANDHARI 54% 5% 8% 33% 0% 0.63 73%
    11 KANDHARI 66% 23% 2% 3% 6% 0.214 64%
    12 KANDHARI 82% 0% 18% 0% 0% 4.476 35%
    13 KANDHARI 57% 29% 14% 0% 0% 3.505 32%
    14 KANDHARI 0% 64% 0% 36% 0% 4.935 29%
    15 KANDHARI 0% 95% 0% 0% 5% 5.14 27%
    16 KANDHARI 47% 53% 0% 0% 0% 5.333 25%
    17 KANDHARI 0% 100% 0% 0% 0% 6.693 24%
    18 KANDHARI 0% 64% 1% 35% 0% 4.929 18%
    19 KANDHARI 0% 94% 6% 0% 0% 6.486 17%
    20 KANDHARI 100% 0% 0% 0% 0% 7.927 16%
    21 KANDHARI 0% 0% 0% 100% 0% 10.573 6%
    22 KANDHARI 0% 0% 100% 0% 0% 106.596 0%
    23 KANDHARI 0% 0% 0% 0% 100% 997.532 0%

  14. The Following 5 Users Say Thank You to Kurd For This Useful Post:

     dp (05-14-2016), Kaido (05-13-2016), khanabadoshi (05-13-2016), Mellifluous (05-14-2016), surbakhunWeesste (05-13-2016)

  15. #8
    Registered Users
    Posts
    4,115
    Sex

    Kandhari's raw results for transparency. Please note that steppe ancestry is transmitted via either Andronovo or Scythian IA. The Andronovo sample used was the higher coverage 503 sample.

    0 .Kandhari 1
    1 .Kurd_C3 1
    2 Andronovo2 1
    3 Palliyar 5
    4 Scythian_IA 1
    5 Mongola 6
    6 MbutiPygmy 10
    7 Karitiana 12
    8 Onge 9
    9 Han 33
    10 Yoruba 70
    11 Ami 10
    jackknife block size: 0.05
    snps: 532403 indivs: 159
    number of blocks for block jackknife: 711
    dof (jackknife): 617.222
    numsnps used: 26200
    codimension 1
    f4info:
    f4rank: 4 dof: 1 chisq: 0.214 tail: 0.643540517 dofdiff: 3 chisqdiff: -0.214 taildiff: 1
    B:
    scale 1 1 1 1
    Karitiana 0.833 -0.931 1.829 -0.369
    Onge 0.685 1.979 0.704 -0.108
    Han 1.41 -0.231 -0.787 -1.37
    Yoruba -0.03 -0.351 -0.005 1.363
    Ami 1.36 -0.204 -0.735 1.057
    A:
    scale 147.93 1063.562 2185.674 9230.83
    .Kurd_C3 -0.247 -0.024 -0.643 0.185
    Andronovo2 0.021 0.04 1.755 -0.869
    Palliyar 0.468 2.168 0.184 0.557
    Scythian_IA 0.36 -0.388 1.191 1.974
    Mongola 2.142 -0.382 -0.235 -0.048


    full rank 1
    f4info:
    f4rank: 5 dof: 0 chisq: 0 tail: 1 dofdiff: 1 chisqdiff: 0.214 taildiff: 0.643540517
    B:
    scale 1 1 1 1 1
    Karitiana 0.833 -0.928 1.853 -0.1 -0.008
    Onge 0.684 1.977 0.695 0.202 0.319
    Han 1.405 -0.244 -0.811 -1.149 0.994
    Yoruba -0.036 -0.372 -0.083 1.494 1.619
    Ami 1.365 -0.182 -0.646 1.182 -1.135
    A:
    scale 148.874 1068.715 2249.368 9789.623 18729.193
    .Kurd_C3 -0.273 -0.053 -0.769 -0.265 -2.064
    Andronovo2 0.002 0.026 1.736 -1.325 -0.478
    Palliyar 0.456 2.161 0.119 0.269 -0.195
    Scythian_IA 0.342 -0.41 1.134 1.723 -0.678
    Mongola 2.145 -0.4 -0.305 -0.365 -0.112


    best coefficients: 0.658 0.225 0.02 0.033 0.064
    ssres:
    -0.000174082 -0.000208763 -0.000194681 -0.000178363 -0.000129663
    -0.971896434 -1.165515148 -1.086897391 -0.995796655 -0.723902511

    Jackknife mean: 0.641518114 0.163045599 0.039419521 0.108697179 0.047319587
    std. errors: 0.3 0.432 0.145 0.522 0.105

    error covariance (* 1000000)
    90076 -4175 -15723 -89536 19359
    -4175 186227 -37541 -175652 31141
    -15723 -37541 21065 43045 -10847
    -89536 -175652 43045 272761 -50618
    19359 31141 -10847 -50618 10965


    fixed pat wt dof chisq tail prob
    0 0 1 0.214 0.643541 0.658 0.225 0.02 0.033 0.064
    1 1 2 0.63 0.729771 0.541 0.051 0.08 0.328 0
    10 1 2 0.219 0.896408 0.541 0.051 0.08 0.328 0
    100 1 2 0.234 0 0.673 0.259 0 -0.005 0.074 infeasible
    1000 1 2 0.53 0.767192 0.672 0 0.063 0.236 0.029
    10000 1 2 2.338 0 0 -2.297 0.587 3.247 -0.537 infeasible
    11 2 3 3.505 0.320167 0.573 0.29 0.137 0 0
    101 2 3 1.209 0.750881 0.517 0.093 0 0.39 0
    110 2 3 0.235 0.971825 0.671 0.256 0 0 0.073
    1001 2 3 0.657 0.883358 0.572 0 0.084 0.345 0
    1010 2 3 0.973 0.807725 0.883 0 0.04 0 0.077
    1100 2 3 0.802 0.849027 0.76 0 0 0.188 0.052
    10001 2 3 4.929 0.17706 0 0.636 0.009 0.354 0
    10010 2 3 5.055 0 0 0.99 -0.048 0 0.058 infeasible
    10100 2 3 3.725 0 0 -5.176 0 7.185 -1.009 infeasible
    11000 2 3 3.186 0 0 0 0.143 0.984 -0.127 infeasible
    111 3 4 5.333 0.254829 0.468 0.532 0 0 0
    1011 3 4 4.476 0.345384 0.823 0 0.177 0 0
    1101 3 4 1.291 0.862965 0.575 0 0 0.425 0
    1110 3 4 1.077 0.897923 0.914 0 0 0 0.086
    10011 3 4 6.486 0.16566 0 0.94 0.06 0 0
    10101 3 4 4.935 0.294031 0 0.638 0 0.362 0
    10110 3 4 5.14 0.273193 0 0.951 0 0 0.049
    11001 3 4 10.555 0 0 0 -0.02 1.02 0 infeasible
    11010 3 4 98.443 0 0 0 1.098 0 -0.098 infeasible
    11100 3 4 4.034 0 0 0 0 1.121 -0.121 infeasible
    1111 4 5 7.927 0.160292 1 0 0 0 0
    10111 4 5 6.693 0.244483 0 1 0 0 0
    11011 4 5 106.596 2.15E-21 0 0 1 0 0
    11101 4 5 10.573 0.0605296 0 0 0 1 0
    11110 4 5 997.532 0 0 0 0 0 1
    best pat: 0 0.643541 - -
    best pat: 10 0.896408 chi(nested): 0.005 p-value for nested model: 0.946063
    best pat: 110 0.971825 chi(nested): 0.016 p-value for nested model: 0.899809
    best pat: 1110 0.897923 chi(nested): 0.842 p-value for nested model: 0.358737
    best pat: 10111 0.244483 chi(nested): 5.616 p-value for nested model: 0.0177949

  16. The Following User Says Thank You to Kurd For This Useful Post:

     surbakhunWeesste (05-13-2016)

  17. #9
    Gold Member Class
    Posts
    848
    Sex
    Location
    event horizon
    Ethnicity
    Ibex retter; Der Nomad
    Nationality
    Kochyan
    Y-DNA
    R1a......(Dad)
    mtDNA
    U2b2

    Afghanistan United States of America
    Wrorajana, dagha pezrrepory de, tase bel kavkav naoomne laeri?… eshtamaal Weishei!

    Quote Originally Posted by DMXX View Post
    Iranians and Kurds contain some steppe admixture themselves, so using them as a surrogate for more recent West Asian ancestry in Pashtuns would, in turn, cause a depreciation in their actual steppe ancestry. One workaround were if you used ADMIXTURE to identify those Iranians and Kurds with the least steppe admixture and utilise those?
    Last edited by surbakhunWeesste; 05-13-2016 at 10:03 PM.

  18. The Following User Says Thank You to surbakhunWeesste For This Useful Post:

     Mellifluous (05-14-2016)

  19. #10
    Registered Users
    Posts
    4,115
    Sex

    Quote Originally Posted by DMXX View Post
    The "Scythian_IA" sample comes from the Volga region; odds are it will not resemble the Iron Age Saka found in places like Tajikistan. Going by the GD's in several calculators (punt K12 Ancient below), they look like some sort of mixture between the older Yamnaya-Afanasievo era and more recent Sintashta-Andronovo era steppe.

     

    # Population (source) Distance
    1 Scythian_IA_I0247 0.86
    2 Yamnaya_Samara_I0443 12.52
    3 Andronovo_SG_RISE505 13.05
    4 Poltavka_I0440 13.81
    5 Yamnaya_Samara_I0231 13.87
    6 Corded_Ware_Germany_I0104 14.17
    7 Yamnaya_Kalmykia_SG_RISE552 14.67
    8 Afanasievo_SG_RISE511 14.83
    9 Srubnaya_I0430 15.87
    10 Sintashta_MBA_RISE_386 15.99


    The physical anthropology indicates the Saka of South-Central Asia were quite different from the populations that resided on the steppes during the Bronze and Iron Age. Craniometry satisfactorily predicted Afanasievo is derived from Yamnaya years before 2015, so this snippet of evidence can't be ruled out so easily.

    If they're being used as a population source in qpAdm for Pashtuns, they'll demonstrate an exaggerated score due to their Yamnaya-type admixture (older steppe samples look more CHG-rich, EHG apparently functions as well as Mal'ta for replicating ANE admixture). In retrospect, this would explain why nearly everyone from Kurdistan to the Subcontinent has a smaller GD with Scythian_IA than the actual steppe source of proto-Indo-Iranian in Asia (Sintashta) or its' direct derivatives (Andronovo).

    Iranians and Kurds contain some steppe admixture themselves, so using them as a surrogate for more recent West Asian ancestry in Pashtuns would, in turn, cause a depreciation in their actual steppe ancestry. One workaround were if you used ADMIXTURE to identify those Iranians and Kurds with the least steppe admixture and utilise those?

    Another option might be to consider using Kotias, Sintashta (or the least East Eurasian Andronovo samples if deamination is an issue), Iraqi/Iranian Jew, Pulliyar and Korean? Depends what the aspirations are.

    [Edit]: Just read your post above, that second option probably doesn't serve the goal in that case.
    The program does the picking and assigns the fit. So the results do show scenarios with lower or 0% Kurd, with all the caucuses ancestry inferred via Scythian and Androvo. However, the fits are not as good. I guess it boils down to a timeline. Two questions need to be answered:

    1- When did the caucauses based groups, whether Kurd, proto-Kurd. or other migrate to SC Asia
    2- How much Andronovo or Scythian related admixture did they possess when they migrated. We have to remember that even if they had not received any steppe related from an Andronovo decendant loopback, they would still have shared common ancestry with Andronovo or Scythian since both were substantially Caucausus area derived. In either case, whether they had geneflow from a Scythian or Andronovo descendant loopback, or not, there presence in SC Asia, and in the run would go to reduce Andronovo or Scythian proportions.

    Yes, you are correct the Scythian sample is from the Volga, and likely less E Eurasian shifted. I suppose with a more E Eurasian shifted sample, the program may reduce the Scythian % as to not overload the Pashtun sample with E Eurasian, and if this occurs more steppe would have to be assigned to Andronovo to compensate.

    The problem with figuring out how much Andronovo derived + Andronovo base admixture the W Asians in SC Asia had, is that the migrations from W Asia may have occurred continuously throughout the neolithic, the bronze age, and later (ie Baloch type groups from within the past 1500 years)

    EDIT: The idea of using Iranian Jews or Iraqi Jews is is not a bad one. Not sure how good or bad the fits will be. I'll give it a try later
    Last edited by Kurd; 05-13-2016 at 10:53 PM.

  20. The Following User Says Thank You to Kurd For This Useful Post:

     Mellifluous (05-14-2016)

Page 1 of 6 123 ... LastLast

Similar Threads

  1. Replies: 88
    Last Post: 07-17-2017, 09:18 PM
  2. Replies: 152
    Last Post: 09-29-2016, 06:15 AM
  3. What is the difference between Afghan and Pashtun?
    By Alanson in forum General Sociology/Ethnology
    Replies: 12
    Last Post: 08-03-2016, 08:17 PM
  4. Replies: 18
    Last Post: 04-22-2015, 01:04 PM
  5. At Least 3 Deep Lineages in the Native American Population
    By AppalachianGumbo in forum American (North)
    Replies: 2
    Last Post: 11-01-2012, 01:23 AM

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •