Page 1 of 17 12311 ... LastLast
Results 1 to 10 of 166

Thread: Isolating pontic steppe admixture in south asians in nmonte

  1. #1
    Bronze Class Member
    Posts
    2,218
    Sex
    Location
    Krasnoyarsk

    Siberian Tatars Tajikistan

    Isolating pontic steppe admixture in south asians in nmonte

    I created this thread because the topic comes up so frequently that its worth having a discussion in its own right. We often struggle to do this when modeling so it might be worth coming up with ideas as to how to do it.

    One model that seems to work well combines ancients and moderns. It is a compact Paniya, Velamas, Krasnoyarsk(or Sintashta), Dzharkutan1, Naxi model. I wanted to use mostly moderns but there is no steppe deficient modern population that is also high Iran_N, so I added Dzharkutan.
    Krasnoyarsk instead of Sintashta to capture slight excess Siberian. It will work for people who are fully indigenous to the region so if you have a foreign ancestor you may need to add something else in.

    Default penalty:
    steppe run.jpg

    The fits are decent but more importantly, we see the steppe values we expect. Midi scores .8-1.67% each run(most likely just noise), Tip and Kush both get zero, and importantly they get little to no BMAC, as expected. Meanwhile my friend aaronbee gets about 26% steppe which is also reasonable.

    p.s. im aware the dzarkutan for balochis is inflated
    Last edited by Censored; 07-26-2019 at 08:38 PM.

  2. The Following 9 Users Say Thank You to Censored For This Useful Post:

     26284729292 (07-27-2019),  Agamemnon (07-27-2019),  Awale (08-07-2019),  DMXX (07-27-2019),  FrostAssassin0701 (07-26-2019),  Kulin (07-29-2019),  kush (08-05-2019),  Lollybolly (07-26-2019),  pegasus (07-27-2019)

  3. #2
    Registered Users
    Posts
    196
    Nationality
    Portugal, France & Canada

    "sample": "Custom:AGUser_Lollybolly",
    "fit": 2.5934,
    "UZB_Dzharkutan1_BA": 47.5,
    "Velamas": 20.83,
    "RUS_Krasnoyarsk_MLBA": 16.67,
    "Paniya": 15,

    The fit isn't that great on me

  4. The Following User Says Thank You to Lollybolly For This Useful Post:

     Censored (07-26-2019)

  5. #3
    Bronze Class Member
    Posts
    2,218
    Sex
    Location
    Krasnoyarsk

    Siberian Tatars Tajikistan
    Some more runs with more users and groups:
    steppe run 2.jpg

    Quote Originally Posted by Lollybolly View Post
    "sample": "Custom:AGUser_Lollybolly",
    "fit": 2.5934,
    "UZB_Dzharkutan1_BA": 47.5,
    "Velamas": 20.83,
    "RUS_Krasnoyarsk_MLBA": 16.67,
    "Paniya": 15,

    The fit isn't that great on me
    It's not meant to be very good, just good enough while serving the purpose.

  6. The Following User Says Thank You to Censored For This Useful Post:

     FrostAssassin0701 (07-26-2019)

  7. #4
    Gold Class Member
    Posts
    1,756
    Sex
    Location
    NorCal
    Ethnicity
    Jatt Sikh
    Nationality
    American & Canadian
    mtDNA (M)
    HV2a3 / R0
    Y-DNA (P)
    L1a2a1 / L1a1

    United States of America California Republic Canada India Punjab Sikh Empire Nishan Sahib
    If there isn't a huge focus on getting fits down, is there any reason to mix moderns with ancients? Can't we use SIS BA3 instead of Velamas and Paniya? Those two will cause some variation in the amount of BMAC scored as well since Paniyas are so much more AASI shifted than Velama who in turn are much more Iran N shifted than Paniya.

    With regards to SIS BA3, the only issue would be very AASI shifted individuals who are more AASI than SISBA3. Also, I'm curious about why Dzharkutan1 was selected versus Gonur1 BA or any other BMAC pop for that matter? Did you just select it randomly?
    Last edited by Sapporo; 07-26-2019 at 10:22 PM.
    pegasus modeling:

    sample": "Punjabi_Jat:Sapporo_AGUser",
    "fit": 1.1506,
    "IRN_Shahr_I_Sokhta_BA3": 43.33,
    "TKM_Gonur1_BA": 31.67,
    "RUS_Sintashta_MLBA": 25,
    "closestDistances": [

  8. The Following 3 Users Say Thank You to Sapporo For This Useful Post:

     aaronbee2010 (07-26-2019),  DMXX (07-27-2019),  MonkeyDLuffy (07-26-2019)

  9. #5
    Bronze Class Member
    Posts
    2,218
    Sex
    Location
    Krasnoyarsk

    Siberian Tatars Tajikistan
    Quote Originally Posted by Sapporo View Post
    If there isn't a huge focus on getting fits down, is there any reason to mix moderns with ancients? Can't we use SIS BA3 instead of Velamas and Paniya? Those two will cause some variation in the amount of BMAC scored as well since Paniyas are so much more AASI shifted than Velama who in turn are much more Iran N shifted than Paniya.

    With regards to SIS BA3, the only issue would be very AASI shifted individuals who are more AASI than SISBA3. Also, I'm curious about why Dzharkutan1 was selected versus Gonur1 BA or any other BMAC pop for that matter? Did you just select it randomly?
    This is mainly for checking steppe admixture. Dzharkutan and gonur1 are very similar, I tried both of them and there was no difference in fits. I felt its understood that using one would mean we could use the other.

    I normally dont mix ancient and modern samples but sometimes it is helpful for finding out certain ancestries. The reason is Paniya+Velama would already contain a lot of AASI,WSHG, Iran_N, Barcin, etc. mixtures which ancients may lack so we can have a simpler model(which IMO is better). SIS3 lacks barcin for example. In any case there is a lack of relevant ancients for many groups.

    I think at some point people did a Paniya+ Tajik Rushan+Velama model which also works in many cases but it means that Tajik would be rather inflated for anyone with lower AASI than Velama, even if they don't have much steppe. The only way to fix this is to add an Iran_N rich ancient and then we would be dealing with just one ancient instead of 2. Is that something youre interested in?

  10. The Following User Says Thank You to Censored For This Useful Post:

     26284729292 (07-27-2019)

  11. #6
    Gold Class Member
    Posts
    1,756
    Sex
    Location
    NorCal
    Ethnicity
    Jatt Sikh
    Nationality
    American & Canadian
    mtDNA (M)
    HV2a3 / R0
    Y-DNA (P)
    L1a2a1 / L1a1

    United States of America California Republic Canada India Punjab Sikh Empire Nishan Sahib
    Quote Originally Posted by Censored View Post
    I normally dont mix ancient and modern samples but sometimes it is helpful for finding out certain ancestries. The reason is Paniya+Velama would already contain a lot of AASI,WSHG, Iran_N, Barcin, etc. mixtures which ancients may lack so we can have a simpler model(which IMO is better). SIS3 lacks barcin for example. In any case there is a lack of relevant ancients for many groups.
    I think mixing moderns with ancients can be useful when looking for trends and clines in ancestry but including them both in a model that includes SISBA3 seems superfluous. With regards to ANF (Barcin N), there is already a notable amount in BMAC pops such as Dzharkutan1 and Gonur1 BA I believe. Both have around 15% I believe.

    I think at some point people did a Paniya+ Tajik Rushan+Velama model which also works in many cases but it means that Tajik would be rather inflated for anyone with lower AASI than Velama, even if they don't have much steppe. The only way to fix this is to add an Iran_N rich ancient and then we would be dealing with just one ancient instead of 2. Is that something youre interested in?

    The issue I have always had with the model is that it separates BMAC into a combo of Tajik Rushan and Velama when BMAC pops are fairly distinct from both. It also makes no sense archaeology wise. An IVC like pop such as the Velama didn't mix directly with a Proto Indo-Aryan like Tajik Rushan pop directly. It works as a basic Steppe + IVC + additional AASI/low Iran N (for those who need it) model though. I'm not looking to do that necessarily. I just don't see they need to add moderns to isolate Pontic Steppe. The standard SISBA3 + Pontic Steppe + BMAC pop works for most individuals. The only issue is very AASI shifted individuals. For them, we are missing a very AASI shifted (Paniya-like) ancient. I would be fine with adding simulated AASI in that case if it didn't cause BMAC proportions to go out of wack when added.
    Last edited by Sapporo; 07-27-2019 at 12:04 AM.
    pegasus modeling:

    sample": "Punjabi_Jat:Sapporo_AGUser",
    "fit": 1.1506,
    "IRN_Shahr_I_Sokhta_BA3": 43.33,
    "TKM_Gonur1_BA": 31.67,
    "RUS_Sintashta_MLBA": 25,
    "closestDistances": [

  12. The Following 2 Users Say Thank You to Sapporo For This Useful Post:

     DMXX (07-27-2019),  MonkeyDLuffy (07-26-2019)

  13. #7
    Bronze Class Member
    Posts
    2,218
    Sex
    Location
    Krasnoyarsk

    Siberian Tatars Tajikistan
    Disclaimer btw, Im not promoting this as a standard model, at all if that's what you're thinking. Im just saying this is a good way to cross-check for steppe the same way that using simulated AASI can be used to check what groups are appropriate. Samples living around the same time period are always the go to as long as that's possible.

    Mixing them together can be ok if you are very careful with it, in this case I only had one category for steppe, one for BMAC. Many people do require a higher AASI population and I would be hesitant to just use SIS3 alone anyway, SSo seems to be the far better option for most folks. This is a good ancient model but it still requires a tribal pop until we get aDNA.

    penalty default
    steppe run 3.jpg

    Even West Siberia could be taken out lol, it would work fine w/o it.

  14. The Following 2 Users Say Thank You to Censored For This Useful Post:

     26284729292 (07-27-2019),  FrostAssassin0701 (07-26-2019)

  15. #8
    Gold Class Member
    Posts
    1,756
    Sex
    Location
    NorCal
    Ethnicity
    Jatt Sikh
    Nationality
    American & Canadian
    mtDNA (M)
    HV2a3 / R0
    Y-DNA (P)
    L1a2a1 / L1a1

    United States of America California Republic Canada India Punjab Sikh Empire Nishan Sahib
    Quote Originally Posted by Censored View Post
    Disclaimer btw, Im not promoting this as a standard model, at all if that's what you're thinking. Im just saying this is a good way to cross-check for steppe the same way that using simulated AASI can be used to check what groups are appropriate. Samples living around the same time period are always the go to as long as that's possible.

    Mixing them together can be ok if you are very careful with it, in this case I only had one category for steppe, one for BMAC. Many people do require a higher AASI population and I would be hesitant to just use SIS3 alone anyway, SSo seems to be the far better option for most folks. This is a good ancient model but it still requires a tribal pop until we get aDNA.

    penalty default
    steppe run 3.jpg

    Even West Siberia could be taken out lol, it would work fine w/o it.
    I understand what you're doing. I just don't think it's a more accurate model to isolate Pontic Steppe than the standard SISBA3 + Pontic Steppe pop + BMAC pop. As for SSo being a far better option, that's the case for many Gangetic populations or those with less of an Iran N pull but for others, the difference is negligible or SISBA3 is preferred. Regardless, SSo is not an ideal ancient to isolate Pontic Steppe ancestry since it is part Pontic Steppe itself. Somewhere in the vicinity of 10% I believe.

    Regarding West Siberia N, I sometimes question why it is added to models even when it makes a negligible improvement to the fit or in some cases, it even worsens it since there is enough Siberian admixture accounted for in other ancients (such as Sarazm En). It's the same case for Chokhopani which is needed for some groups such as Khas Brahmins or Pashtuns with minor East Asian admixture but tends to worsen or not improve fits for others.

    The model you are using above is definitely inflating BMAC proportions due to the inclusion of both Irula + SSo. Both are AASI heavy and much lower Iran N than most of the individuals you are modeling. Try running that model without Irula and a variation that uses SISBA3 instead of SSo for a more accurate Pontic Steppe estimate.

    The model I've been trying to use for isolating Pontic Steppe is SISBA3 + Krasnoyarsk + Sarazm En (for overflow of West Siberia) + Gonur1 BA. This works for NWers obviously but something additional is needed for individuals who are Iran N deficient (otherwise using SSo will eat up some Steppe) or individuals who are more AASI than SIS BA3.
    Last edited by Sapporo; 07-27-2019 at 12:39 AM.
    pegasus modeling:

    sample": "Punjabi_Jat:Sapporo_AGUser",
    "fit": 1.1506,
    "IRN_Shahr_I_Sokhta_BA3": 43.33,
    "TKM_Gonur1_BA": 31.67,
    "RUS_Sintashta_MLBA": 25,
    "closestDistances": [

  16. #9
    Registered Users
    Posts
    813
    Sex

    Quote Originally Posted by Censored View Post
    Disclaimer btw, Im not promoting this as a standard model, at all if that's what you're thinking. Im just saying this is a good way to cross-check for steppe the same way that using simulated AASI can be used to check what groups are appropriate. Samples living around the same time period are always the go to as long as that's possible.

    Mixing them together can be ok if you are very careful with it, in this case I only had one category for steppe, one for BMAC. Many people do require a higher AASI population and I would be hesitant to just use SIS3 alone anyway, SSo seems to be the far better option for most folks. This is a good ancient model but it still requires a tribal pop until we get aDNA.

    penalty default
    steppe run 3.jpg

    Even West Siberia could be taken out lol, it would work fine w/o it.
    Removing Irula, WSHG, and adding Chokhopani made my fit better:

    "sample": "Brahmin_Khas_Nepal:pnb_AGUser",
    "fit": 2.7278,
    "PAK_Swat_Saidu_Sharif_IA_o": 53.33,
    "UZB_Dzharkutan1_BA": 25,
    "RUS_Sintashta_MLBA": 15,
    "NPL_Chokhopani_2700BP": 6.67,

  17. The Following User Says Thank You to pnb123 For This Useful Post:

     bmoney (08-05-2019)

  18. #10
    Gold Class Member
    Posts
    1,756
    Sex
    Location
    NorCal
    Ethnicity
    Jatt Sikh
    Nationality
    American & Canadian
    mtDNA (M)
    HV2a3 / R0
    Y-DNA (P)
    L1a2a1 / L1a1

    United States of America California Republic Canada India Punjab Sikh Empire Nishan Sahib
    Quote Originally Posted by pnb123 View Post
    Removing Irula, WSHG, and adding Chokhopani made my fit better:

    "sample": "Brahmin_Khas_Nepal:pnb_AGUser",
    "fit": 2.7278,
    "PAK_Swat_Saidu_Sharif_IA_o": 53.33,
    "UZB_Dzharkutan1_BA": 25,
    "RUS_Sintashta_MLBA": 15,
    "NPL_Chokhopani_2700BP": 6.67,
    Try swapping around Gonur1 BA and SISBA3 with SSo and Dzharkutan1_BA as well. Krasnoyarsk tends to improve fits universally in South Asia as well since it's a little more CHG/WSHG shifted than Sintashta. I'm sure you could get the fit below 2.5 using those 3 in some manner with NPL_Chokhopani_2700BP.
    pegasus modeling:

    sample": "Punjabi_Jat:Sapporo_AGUser",
    "fit": 1.1506,
    "IRN_Shahr_I_Sokhta_BA3": 43.33,
    "TKM_Gonur1_BA": 31.67,
    "RUS_Sintashta_MLBA": 25,
    "closestDistances": [

  19. The Following User Says Thank You to Sapporo For This Useful Post:

     pnb123 (07-27-2019)

Page 1 of 17 12311 ... LastLast

Similar Threads

  1. Neanderthal Admixture in South Asians
    By traject in forum Southern
    Replies: 9
    Last Post: 03-29-2019, 02:42 PM
  2. Replies: 11
    Last Post: 02-17-2019, 08:54 AM
  3. Replies: 2
    Last Post: 05-15-2018, 08:55 PM
  4. Steppe Ancestry in S & W Asians
    By Kurd in forum Ancient (aDNA)
    Replies: 147
    Last Post: 03-27-2016, 03:49 AM
  5. Bronze Age Steppe and European mtdna in South Asians
    By Raskolnikov in forum Ancient (aDNA)
    Replies: 1
    Last Post: 07-01-2015, 03:03 PM

Posting Permissions

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