Page 1 of 4 123 ... LastLast
Results 1 to 10 of 36

Thread: No, Steppe_MLBA did not give ancestry to Swat valley iron age

  1. #1

    No, Steppe_MLBA did not give ancestry to Swat valley iron age

    What is the source of the steppe ancestry in Swat Valley IA (SPGT)?

    Indus Periphery + Steppe_mlba(either western or central) is a massive fail. Narsimhan paper has extremely poor p-values for his qpAdm model with left pops (SPGT, Western_steppe_mlba/central_steppe_mlba, Onge) - 0.003 and 0.006 (FigS50, Pg 283, Narsimhan 2019 Supplement). He makes the 0.006 pass and justifies it (imo) using statistical gimmickry (Bonferroni correction). He does not even follow his own threshold as shown in the quote from Pg187-188 from Narsimhan 2019 supplement file. To be fair, he does give the caveat that mass testing may pass many bad models, which is what happened in this case. Emphasis in below quote is mine.

    We were aware that due to the large number of tests we carried out, we might find models that pass simply as a statistical fluctuation even if the truth is that the models do not fit (the problem of multiple hypothesis testing). However, it is important to recognize that due to shared drift between populations, many of our tested models for admixture proportions are not independent, and so the number of tested models is not effectively as large as it seems.In addition, multiple hypothesis testing is not as problematic for our procedure as it might at first seem because our process is a test of acceptance of a model and not rejection, and thus increasing the number of tests should decrease the rate of observing feasible working models at a fixed p-value threshold. We use a p-value at 0.01 as acceptance for the interpretation of working models, a level that is far more conservative than what we would obtain by Bonferroni correction (110-8)

    That model also creates 3 very pertinent unanswered questions
    1. Are we supposed to believe that Sintashta_mlba like population jumped over BMAC without mixing and came into swat?
    2. Why so much dearth of R1a in swat if sintashta_mlba was 100pc R1a?
    3. Where did the Levant y haplogroup E1b1b1b2 (2nd highest freq in swat) come from?


    I think i have the answer.

    In the qpAdm i have removed 1 sample from the Indus periphery 11 because it is an outlier. I8728 from SiS dated 2500-2600bce has a huge amount of AASI compared to the 10 others. This person likely was a migrant from the comparatively southern (but bordering modern day pakistan) region of Gujarat and surrounds.

    • InPe - 10 samples labeled Indus Periphery excluding 11th sample I8728 which is an AASI heavy outlier.
    • Armenia_MBA are 4 samples from 1900-1200bce. 1 of which is E1b1b1b2a.
    • Molaly_LBA is from east Kazakshtan close to Tarim basin and Dali area, arch dated to 1400-1000bce. Its ancestry is unique, steppe_mlba shifted towards Shamanka_N (east asian), minor Botai and heavily towards south asia.
    • SPGT - 85 samples from Swat valley IA (labelled Loebanr_IA, Katelai_IA, Udegram_IA, Gogdara_IA, Arkotkila_IA, Burkara_IA, Barikot_IA and tested together by Narsimhan, like I have done below)


    qpAdm Output

    SPGT
    InPe - 0.597+-0.022
    Armenia_MBA - 0.124 +-0.020
    Molaly_LBA - 0.279 +-0.028
    chisq 12.64 tail prob 0.244
    Output file
    Model passes test, so SPGT can be modeled as 60% Indus periphery + 12% Armenia_MBA + 28% Molaly_LBA

    Sintashta_MLBA instead of Molaly_LBA is a big fail with p-value in power of e-5.
    Output file

    Oy_Dzhaylau_MLBA is from the east Kazakh area. It is also a big fail in place of Molaly.
    Output file

    Dali_MBA instead of Molaly gives p-value of 0.037, better than the above 2, but not great.
    Output file

    It is clear from qpAdm output gendstats that SPGT requires some east asian (Shamanka_N related) as well as Botai related ancestry, which Molaly provides, but not any other steppe_mlba source.

    In summation:
    1. Armenia_MBA type ancestry came to Swat between 1900-1200bce, likely in the early half of 2nd mill bce. Brought the Levant E1b1b1b2 y haplogroup to Swat.
    2. Next, Molaly type LBA steppe ancestry came to Swat through inner asian mountain corridor post 1400bce. This is the reason why BMAC has little admixture from the steppe, as latest bmac samples in Narsimhan paper are only till 1300-1400bce. This ancestry likely brought the couple of Q1a and Q1b samples we see in SPGT.
    3. The idea that a steppe_mlba ancestry came to Swat valley between 2000-1500bce after jumping/sidestepping BMAC is BS.
    Last edited by misnomer; 01-10-2020 at 12:06 AM.

  2. The Following 3 Users Say Thank You to misnomer For This Useful Post:

     laltota (01-15-2020),  palamede (01-10-2020),  tipirneni (01-12-2020)

  3. #2
    Registered Users
    Posts
    4,612
    Sex

    Indians are still related to Balts linguistically and genetically, and the genetic connection is obviously Steppe_MLBA.

    So Indo-Iranian languages arrived in South Asia via Steppe_MLBA, and nothing really changes.
    Last edited by Generalissimo; 01-09-2020 at 11:44 PM.

  4. The Following 6 Users Say Thank You to Generalissimo For This Useful Post:

     Hando (01-10-2020),  laltota (01-15-2020),  palamede (01-10-2020),  peternorth (01-11-2020),  Pribislav (01-10-2020),  sum1 (01-10-2020)

  5. #3
    Registered Users
    Posts
    1,742
    Sex
    Omitted

    I saw something similar a while back...
    left pops:
    Swat_IA
    SIS_BA2 (S8726, S8728, I11456)
    Hasanlu_IA (~43% Kura_Araxes, ~43% Seh_Gabi_CA, ~14% Steppe_MLBA. Overall not drastically different to Armenia_MBA which is mostly Kura_Araxes + Steppe_EBA)
    Kazakhstan_MBA2 (~50% Steppe_MLBA, ~30% Baikal_BA, ~20% Central Asian Neolithic)

    right pops: Barcin_N, Iran_N, CHG, Taforalt, MA1, Brazil_Sumidouro_10100BP, EHG, IGHG_Serbia, Afanasievo, Baikal_EN, Ami.DG, Onge.DG, Ust_Ishim

    SIS_BA2: 0.574 +/- 0.024
    Hasanlu_IA: 0.243 +/- 0.022
    Kazakhstan_MBA2: 0.183 +/- 0.019
    chisq: 11.230 tail prob: 0.339

    Hasanlu_IA works well as a contributor to many modern Central/South-Central Asian populations in quite high proportions.
    I would throw my uneducated guess that would have something to do with middle Iranic languages reaching Central Asia during Persian empire?
    Why related ancestry would be present earlier however?

    Another option worth exploring is Dzharkustan_BA2. We know they were present in Central Asia (whereas a migration of Armenia_MBA would by hypothetical), and are reasonably similar to Armenia_MBA (being that they cluster close to modern Caucasus pops)
    Last edited by Kale; 01-10-2020 at 03:51 AM.
    Collection of 14,000 d-stats: Hidden Content Part 2: Hidden Content Part 3: Hidden Content PM me for d-stats, qpadm, qpgraph, or f3-outgroup nmonte models.

  6. The Following 4 Users Say Thank You to Kale For This Useful Post:

     laltota (01-15-2020),  misnomer (01-10-2020),  palamede (01-10-2020),  tipirneni (01-12-2020)

  7. #4
    Registered Users
    Posts
    944
    Sex

    @misnomer
    Why couldn't the east asian be from sino-tibetans who live right next door?
    You are obviously unwilling to even explore that option as it goes against your narrative.

    Plus there is no excess east Asian that is directly proportional to Steppic ancestry in the rest of the subcontinent, so LBA Steppe is definitely not ancestral to them.

  8. The Following 2 Users Say Thank You to client For This Useful Post:

     aaronbee2010 (01-10-2020),  laltota (01-15-2020)

  9. #5
    Gold Class Member
    Posts
    4,070
    Location
    Shangri La

    Afghanistan Jammu and Kashmir United States of America Canada
    Quote Originally Posted by misnomer View Post
    What is the source of the steppe ancestry in Swat Valley IA (SPGT)?

    Indus Periphery + Steppe_mlba(either western or central) is a massive fail. Narsimhan paper has extremely poor p-values for his qpAdm model with left pops (SPGT, Western_steppe_mlba/central_steppe_mlba, Onge) - 0.003 and 0.006 (FigS50, Pg 283, Narsimhan 2019 Supplement). He makes the 0.006 pass and justifies it (imo) using statistical gimmickry (Bonferroni correction). He does not even follow his own threshold as shown in the quote from Pg187-188 from Narsimhan 2019 supplement file. To be fair, he does give the caveat that mass testing may pass many bad models, which is what happened in this case. Emphasis in below quote is mine.




    That model also creates 3 very pertinent unanswered questions
    1. Are we supposed to believe that Sintashta_mlba like population jumped over BMAC without mixing and came into swat?
    2. Why so much dearth of R1a in swat if sintashta_mlba was 100pc R1a?
    3. Where did the Levant y haplogroup E1b1b1b2 (2nd highest freq in swat) come from?


    I think i have the answer.

    In the qpAdm i have removed 1 sample from the Indus periphery 11 because it is an outlier. I8728 from SiS dated 2500-2600bce has a huge amount of AASI compared to the 10 others. This person likely was a migrant from the comparatively southern (but bordering modern day pakistan) region of Gujarat and surrounds.

    • InPe - 10 samples labeled Indus Periphery excluding 11th sample I8728 which is an AASI heavy outlier.
    • Armenia_MBA are 4 samples from 1900-1200bce. 1 of which is E1b1b1b2a.
    • Molaly_LBA is from east Kazakshtan close to Tarim basin and Dali area, arch dated to 1400-1000bce. Its ancestry is unique, steppe_mlba shifted towards Shamanka_N (east asian), minor Botai and heavily towards south asia.
    • SPGT - 85 samples from Swat valley IA (labelled Loebanr_IA, Katelai_IA, Udegram_IA, Gogdara_IA, Arkotkila_IA, Burkara_IA, Barikot_IA and tested together by Narsimhan, like I have done below)


    qpAdm Output

    SPGT
    InPe - 0.597+-0.022
    Armenia_MBA - 0.124 +-0.020
    Molaly_LBA - 0.279 +-0.028
    chisq 12.64 tail prob 0.244
    Output file
    Model passes test, so SPGT can be modeled as 60% Indus periphery + 12% Armenia_MBA + 28% Molaly_LBA

    Sintashta_MLBA instead of Molaly_LBA is a big fail with p-value in power of e-5.
    Output file

    Oy_Dzhaylau_MLBA is from the east Kazakh area. It is also a big fail in place of Molaly.
    Output file

    Dali_MBA instead of Molaly gives p-value of 0.037, better than the above 2, but not great.
    Output file

    It is clear from qpAdm output gendstats that SPGT requires some east asian (Shamanka_N related) as well as Botai related ancestry, which Molaly provides, but not any other steppe_mlba source.

    In summation:
    1. Armenia_MBA type ancestry came to Swat between 1900-1200bce, likely in the early half of 2nd mill bce. Brought the Levant E1b1b1b2 y haplogroup to Swat.
    2. Next, Molaly type LBA steppe ancestry came to Swat through inner asian mountain corridor post 1400bce. This is the reason why BMAC has little admixture from the steppe, as latest bmac samples in Narsimhan paper are only till 1300-1400bce. This ancestry likely brought the couple of Q1a and Q1b samples we see in SPGT.
    3. The idea that a steppe_mlba ancestry came to Swat valley between 2000-1500bce after jumping/sidestepping BMAC is BS.
    I agree with some of your statements but the E1b in the Swat did not come from Armenia it came from the BMAC and its apparent in many of the uniparental markers of the people there along with Steppe uniparental markers. The ENA in many of them seems localized and is of the Himalayan variety, so no its not being brought from some Botai population because some SPGT samples don't have ENA at all. The earliest samples are from 1200-1300 BC, so using Molaly LBA or Zevakinsky LBA are red herrings, you have to add some Han/ENA related source separately then add a Pontic Steppe source. Also quite a few of the samples also have Afanasievo mixture, likely absorbed by Sintashta related groups moving east and explains the presences of those I2 samples.

    Leaving 8728 makes no sense as a good chunk of the SPGT samples esp at Aligrama and Katelai harbor AASI levels close to those IVC samples, in other words they need a more AASI rich source. Also it has a high SNP count, why would you leave it out? I would use the 4 high SNP samples as your IVC pool ( 8726, 8728, 1459, 1456).


    I reproduced your results here, its a pretty good fit , but from a time and geographical aspect they(Arm EBA and Molaly LBA) are not ancestral at all.


    "sample": "PAK_Udegram_IA:Average",
    "fit": 1.4184,
    "IRN_Shahr_I_Sokhta_BA2": 62.5,
    "KAZ_Molaly_LBA": 23.33,
    "ARM_MBA": 14.17,



    With more proximal sources you get a much better fit and Central Steppe sources are preferred.


    "sample": "PAK_Udegram_IA:Average", (where most E1b is found)
    "fit": 0.8755,
    "IRN_Shahr_I_Sokhta_BA2": 53.33,
    "TKM_Gonur1_BA": 24.17,
    "RUS_Krasnoyarsk_MLBA": 19.17,
    "NPL_Chokhopani_2700BP": 3.33,



    Narsimhan's model is a fail though that is for sure but the spread of Steppe ancestry is definitely spread via Central Steppe MLBA or variations of it with even more WSHG and/or Afanasievo ancestry.

    "sample": "PAK_Udegram_IA:Average",
    "fit": 2.2843,
    "IRN_Shahr_I_Sokhta_BA2": 74.17,
    "CustomGroup_CentralSteppeMLBA": 25.83,
    Last edited by pegasus; 01-10-2020 at 06:44 AM.

  10. The Following 3 Users Say Thank You to pegasus For This Useful Post:

     Hando (01-10-2020),  laltota (01-15-2020),  palamede (01-10-2020)

  11. #6
    Quote Originally Posted by client View Post
    @misnomer
    Why couldn't the east asian be from sino-tibetans who live right next door?
    You are obviously unwilling to even explore that option as it goes against your narrative.

    Plus there is no excess east Asian that is directly proportional to Steppic ancestry in the rest of the subcontinent, so LBA Steppe is definitely not ancestral to them.
    No. Chokhopani type population is definitely not ancestral to SPGT. This is confirmed by adding Chokhopani to the outgroup right pops and seeing that the results do not change for my model (Result file here).

    Chokhopani is shamanka_eba east asian like and not Botai shifted. It is clear from the output of the failed models (like sintashta_mlba, few gendstats from the failed model pasted below) that east asian along with botai is required, not just east asian like chokhopani. Therefore, this ancestry in SPGT definitely came from somewhere in far eastern kazakhstan. I have also not heard of any archaeological evidence that shows migration from nepal/tibet to swat in the bronze age/IA.

    Code:
    gendstat:  Kazakhstan_Botai.SG         Ganj_Dareh_N     4.362
    gendstat:  Kazakhstan_Botai.SG                  PPN     2.292
    gendstat:  Kazakhstan_Botai.SG           Anatolia_N     2.643
    gendstat:  Kazakhstan_Botai.SG            Iberia_MN     2.858
    gendstat:  Kazakhstan_Botai.SG        Russia_MA1.SG     2.885
    gendstat:  Kazakhstan_Botai.SG Georgia_Satsurblia.SG     1.712
    gendstat:  Kazakhstan_Botai.SG        Kostenki14.SG     2.728
    gendstat:  Kazakhstan_Botai.SG    Russia_Karelia_HG     2.611
    gendstat:  Kazakhstan_Botai.SG Serbia_Iron_Gates_HG     4.218
    gendstat:  Kazakhstan_Botai.SG            Tyumen_HG     1.654
    gendstat:  Kazakhstan_Botai.SG              Onge.DG     4.077
    gendstat:  Ganj_Dareh_N            Russia_Shamanka_EN.SG    -3.199
    gendstat:  Onge.DG                Russia_Shamanka_EN.SG    -3.453
    Last edited by misnomer; 01-10-2020 at 09:24 AM.

  12. The Following User Says Thank You to misnomer For This Useful Post:

     palamede (01-10-2020)

  13. #7
    Quote Originally Posted by Kale View Post
    I saw something similar a while back...
    left pops:
    Swat_IA
    SIS_BA2 (S8726, S8728, I11456)
    Hasanlu_IA (~43% Kura_Araxes, ~43% Seh_Gabi_CA, ~14% Steppe_MLBA. Overall not drastically different to Armenia_MBA which is mostly Kura_Araxes + Steppe_EBA)
    Kazakhstan_MBA2 (~50% Steppe_MLBA, ~30% Baikal_BA, ~20% Central Asian Neolithic)

    right pops: Barcin_N, Iran_N, CHG, Taforalt, MA1, Brazil_Sumidouro_10100BP, EHG, IGHG_Serbia, Afanasievo, Baikal_EN, Ami.DG, Onge.DG, Ust_Ishim

    SIS_BA2: 0.574 +/- 0.024
    Hasanlu_IA: 0.243 +/- 0.022
    Kazakhstan_MBA2: 0.183 +/- 0.019
    chisq: 11.230 tail prob: 0.339

    Hasanlu_IA works well as a contributor to many modern Central/South-Central Asian populations in quite high proportions.
    I would throw my uneducated guess that would have something to do with middle Iranic languages reaching Central Asia during Persian empire?
    Why related ancestry would be present earlier however?

    Another option worth exploring is Dzharkustan_BA2. We know they were present in Central Asia (whereas a migration of Armenia_MBA would by hypothetical), and are reasonably similar to Armenia_MBA (being that they cluster close to modern Caucasus pops)
    Quick question. What samples does Kazakhstan_MBA2 represent? I will test the model out. I am comfortable with Armenia_MBA because Narsimhan's ALDER analysis gives mixing date of around 1800-1500bce in SWAT with his model. Another option is some BMAC/turan ancestry instead of Armenia because of the time frame and also presence of E1b1b in I10551 Farkhor,Tajikistan and I2085 Gonur,Turkmenistan. presence of both armenia/south caucasus & turan ancestry apart from Molaly_LBA like ancestry is also a possibility.

  14. #8
    deleted
    Last edited by misnomer; 01-10-2020 at 04:26 PM. Reason: deleted

  15. #9
    Quote Originally Posted by pegasus View Post


    I reproduced your results here, its a pretty good fit , but from a time and geographical aspect they(Arm EBA and Molaly LBA) are not ancestral at all.


    "sample": "PAK_Udegram_IA:Average",
    "fit": 1.4184,
    "IRN_Shahr_I_Sokhta_BA2": 62.5,
    "KAZ_Molaly_LBA": 23.33,
    "ARM_MBA": 14.17,



    With more proximal sources you get a much better fit and Central Steppe sources are preferred.


    "sample": "PAK_Udegram_IA:Average", (where most E1b is found)
    "fit": 0.8755,
    "IRN_Shahr_I_Sokhta_BA2": 53.33,
    "TKM_Gonur1_BA": 24.17,
    "RUS_Krasnoyarsk_MLBA": 19.17,
    "NPL_Chokhopani_2700BP": 3.33,



    Narsimhan's model is a fail though that is for sure but the spread of Steppe ancestry is definitely spread via Central Steppe MLBA or variations of it with even more WSHG and/or Afanasievo ancestry.

    "sample": "PAK_Udegram_IA:Average",
    "fit": 2.2843,
    "IRN_Shahr_I_Sokhta_BA2": 74.17,
    "CustomGroup_CentralSteppeMLBA": 25.83,
    Its good that people agree that Narsimhans model is crude and does not capture reality.

    With that said, you are not giving me anything apart from nMonte runs. That is not formal statistical testing. Once you get such a model, you should run it in qpAdm or qpWave to see if it works. That takes time and patience.

    Your model for Udegram just does not work

    left pops:
    Udegram_IA
    Shahr_I_Sokhta_BA2 - 0.554 +-0.056
    Gonur1_BA - 0.212 +- 0.048
    Krasnoyarsk_MLBA - 0.202 +- 0.014
    Nepal_Chokhopani_2700BP.SG - 0.032 +- 0.012
    chisq 32.11 tailprob 0.0001906

    Output File

    The reason for the failure is also clear. I have already stated it in a post above. Too less Botai.
    Code:
    gendstat:  Kazakhstan_Botai.SG         Ganj_Dareh_N     3.312
    gendstat:  Kazakhstan_Botai.SG                  PPN     3.473
    gendstat:  Kazakhstan_Botai.SG           Anatolia_N     3.651
    gendstat:  Kazakhstan_Botai.SG            Iberia_MN     3.512
    gendstat:  Kazakhstan_Botai.SG        Russia_MA1.SG    -0.304
    gendstat:  Kazakhstan_Botai.SG Georgia_Satsurblia.SG     1.306
    gendstat:  Kazakhstan_Botai.SG        Kostenki14.SG     1.994
    gendstat:  Kazakhstan_Botai.SG    Russia_Karelia_HG     2.263
    gendstat:  Kazakhstan_Botai.SG Serbia_Iron_Gates_HG     4.261
    gendstat:  Kazakhstan_Botai.SG            Tyumen_HG     0.717
    gendstat:  Kazakhstan_Botai.SG              Onge.DG     3.627
    gendstat:  Kazakhstan_Botai.SG Russia_Shamanka_EN.SG     3.283
    As far as BMAC or Gonur being a possible source instead of Armenia_MBA, I tested the models . Theyre ok but not as great as Armenia_MBA. Both just fail my 0.05 tailprob threshold. InPe + Gonur + any western/central MLBA steppe source of your choice is a massive fail. Only Molaly_LBA works. Some results

    left pops:
    SPGT
    InPe - 0.495 +- 0.032
    Gonur1_BA - 0.138 +- 0.027
    Molaly_LBA - 0.367 +- 0.024
    chisq 20.57 tailprob 0.038
    Output File

    left pops:
    SPGT
    InPe - 0.496 +- 0.032
    Bustan_BA - 0.140 +- 0.027
    Molaly_LBA - 0.364 +- 0.024
    chisq 19.73 tailprob 0.0491

    Output file

    However, given the ok results above, it seems likely that there is some minor BMAC admixture in SPGT.

    Quote Originally Posted by pegasus View Post
    Leaving 8728 makes no sense as a good chunk of the SPGT samples esp at Aligrama and Katelai harbor AASI levels close to those IVC samples, in other words they need a more AASI rich source. Also it has a high SNP count, why would you leave it out? I would use the 4 high SNP samples as your IVC pool ( 8726, 8728, 1459, 1456).
    The 10 Indus Periphery samples in my test have 878k SNPs in all. More than sufficient to get accurate results. I8728 is definitely an outlier because the more south easterly Rakhigarhi sample from the same time period is said to have much lesser AHG/AASI.

    Quote Originally Posted by pegasus View Post
    I agree with some of your statements but the E1b in the Swat did not come from Armenia it came from the BMAC and its apparent in many of the uniparental markers of the people there along with Steppe uniparental markers. The ENA in many of them seems localized and is of the Himalayan variety, so no its not being brought from some Botai population because some SPGT samples don't have ENA at all. The earliest samples are from 1200-1300 BC, so using Molaly LBA or Zevakinsky LBA are red herrings, you have to add some Han/ENA related source separately then add a Pontic Steppe source. Also quite a few of the samples also have Afanasievo mixture, likely absorbed by Sintashta related groups moving east and explains the presences of those I2 samples.
    All this means nothing without a working qpAdm model.
    Last edited by misnomer; 01-10-2020 at 04:30 PM.

  16. The Following 2 Users Say Thank You to misnomer For This Useful Post:

     laltota (01-15-2020),  palamede (01-10-2020)

  17. #10
    Gold Class Member
    Posts
    4,070
    Location
    Shangri La

    Afghanistan Jammu and Kashmir United States of America Canada
    Quote Originally Posted by misnomer View Post
    Its good that people agree that Narsimhans model is crude and does not capture reality.

    With that said, you are not giving me anything apart from nMonte runs. That is not formal statistical testing. Once you get such a model, you should run it in qpAdm or qpWave to see if it works. That takes time and patience.

    Your model for Udegram just does not work

    left pops:
    Udegram_IA
    Shahr_I_Sokhta_BA2 - 0.554 +-0.056
    Gonur1_BA - 0.212 +- 0.048
    Krasnoyarsk_MLBA - 0.202 +- 0.014
    Nepal_Chokhopani_2700BP.SG - 0.032 +- 0.012
    chisq 32.11 tailprob 0.0001906

    Output File

    The reason for the failure is also clear. I have already stated it in a post above. Too less Botai.
    Code:
    gendstat:  Kazakhstan_Botai.SG         Ganj_Dareh_N     3.312
    gendstat:  Kazakhstan_Botai.SG                  PPN     3.473
    gendstat:  Kazakhstan_Botai.SG           Anatolia_N     3.651
    gendstat:  Kazakhstan_Botai.SG            Iberia_MN     3.512
    gendstat:  Kazakhstan_Botai.SG        Russia_MA1.SG    -0.304
    gendstat:  Kazakhstan_Botai.SG Georgia_Satsurblia.SG     1.306
    gendstat:  Kazakhstan_Botai.SG        Kostenki14.SG     1.994
    gendstat:  Kazakhstan_Botai.SG    Russia_Karelia_HG     2.263
    gendstat:  Kazakhstan_Botai.SG Serbia_Iron_Gates_HG     4.261
    gendstat:  Kazakhstan_Botai.SG            Tyumen_HG     0.717
    gendstat:  Kazakhstan_Botai.SG              Onge.DG     3.627
    gendstat:  Kazakhstan_Botai.SG Russia_Shamanka_EN.SG     3.283
    As far as BMAC or Gonur being a possible source instead of Armenia_MBA, I tested the models . Theyre ok but not as great as Armenia_MBA. Both just fail my 0.05 tailprob threshold. InPe + Gonur + any western/central MLBA steppe source of your choice is a massive fail. Only Molaly_LBA works. Some results

    left pops:
    SPGT
    InPe - 0.495 +- 0.032
    Gonur1_BA - 0.138 +- 0.027
    Molaly_LBA - 0.367 +- 0.024
    chisq 20.57 tailprob 0.038
    Output File

    left pops:
    SPGT
    InPe - 0.496 +- 0.032
    Bustan_BA - 0.140 +- 0.027
    Molaly_LBA - 0.364 +- 0.024
    chisq 19.73 tailprob 0.0491

    Output file

    However, given the ok results above, it seems likely that there is some minor BMAC admixture in SPGT.



    The 10 Indus Periphery samples in my test have 878k SNPs in all. More than sufficient to get accurate results. I8728 is definitely an outlier because the more south easterly Rakhigarhi sample from the same time period is said to have much lesser AHG/AASI.



    All this means nothing without a working qpAdm model.

    Nmonte mimics qpAdm so thats not really an argument because you have scientists and even David use it. Though the fact Kale and you have some far flung Western Iranian/Caucasus source working points to the fact algorithm is working on a very distal level and shared components between some populations causes the model to fail but from a proximal, archaeological and chronological perspective those populations never drove demic movements in Central Asia.
    Also Molaly in itself has 40% Steppe MLBA and 30% Turan related ancestry, which explains why your Molaly component is so inflated. No population like Armenia EBA made into SC Asia, so its proxying for some Copper Age source, in that case that clear , Tepe Hissar, Geoksiur are proximal. Also SPGT is diverse, some of them have no Steppe and some have 20-25%, so in both cases Molaly would not be useful. Can you model those Katelai I2 samples with Molaly?? Have you tried Afansasievo, Poltvaka? Have you also tried more proximal WSHG sources like Kumsay and Aigyrzhal?? If you can utilize combinations of these samples , it would be interesting.

    "sample": "KAZ_Molaly_LBA:Average",
    "fit": 1.8156,
    "RUS_Sintashta_MLBA": 38.33,
    "TKM_Gonur1_BA": 30,

    "Naxi": 15.83,
    "RUS_Tyumen_HG": 15.83,
    Last edited by pegasus; 01-11-2020 at 05:39 AM.

  18. The Following 2 Users Say Thank You to pegasus For This Useful Post:

     laltota (01-15-2020),  palamede (01-12-2020)

Page 1 of 4 123 ... LastLast

Similar Threads

  1. Modelling Iron age ancestry
    By Nino90 in forum Autosomal (auDNA)
    Replies: 86
    Last Post: 07-18-2020, 03:18 PM
  2. Replies: 9
    Last Post: 12-04-2019, 06:17 AM
  3. Replies: 3
    Last Post: 04-06-2019, 08:35 PM
  4. besides ancestry what does dna land give you
    By SlytherclawRavenpuff in forum DNA.Land
    Replies: 5
    Last Post: 02-11-2018, 10:50 PM
  5. Replies: 282
    Last Post: 08-01-2017, 09:58 PM

Tags for this Thread

Posting Permissions

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