Page 2 of 14 FirstFirst 123412 ... LastLast
Results 11 to 20 of 140

Thread: Narasimhan values

  1. #11
    Gold Class Member
    Posts
    1,755
    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 Lollybolly View Post
    The data is very strange. I'd highly doubt that Lohanas are the least "Onge" shifted amongst all other Indic groups. Plus, how are Lohanas scoring more steppe than Jatts and twice as low "Onge" than Jatts?
    The model itself is wrong as it ignores the BMAC contribution to South Asia. It's been discussed ad nauseam before but I've pointed out that there is likely a combination of mislabeling and/or sampling issues (ie strong sampling of outliers) of the Jatt samples. This also occurred in the later Daamgard paper and we confirmed it with the few outliers among the Khatri and Kamboj samples, which were publicly available. The difference is that with Jatt Sikhs in Punjab is they are a much more numerous population (than either Khatris or Kamboj) and it is common for non Jatts/mixed individuals in Punjab to identify as such/use their surnames for perceived socio-cultural gain.

    If the Narasimhan samples become available, we can confirm it as well by running them through GEDMatch.
    Last edited by Sapporo; 08-08-2019 at 01:51 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": [

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

     Censored (08-08-2019),  Kulin (08-08-2019),  laltota (08-08-2019),  MonkeyDLuffy (08-08-2019)

  3. #12
    Registered Users
    Posts
    2,773
    Location
    Gonur Tepe

    Afghanistan Jammu and Kashmir United States of America Canada
    Quote Originally Posted by agent_lime View Post
    I agree. I am just saying the model is absurd. Did you add more Iran_N in the simulation IVCp?
    no IVCp in itself has 70% Iran_N related ancestry. Though I don't think those Iran_N rich groups are going to be needed at all for Gangetic populations.

    I made a post on that here with some pics

    https://anthrogenica.com/showthread....l=1#post589620

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

     bmoney (08-13-2019)

  5. #13
    Administrator
    Posts
    3,688
    Sex
    mtDNA (M)
    D4j5*
    Y-DNA (P)
    R2a*-M124 (L295-)

    England
    I re-read Anthony's The Horse, Wheel and Language last night regarding the Indo-Aryans. See here for more.

    Just to re-affirm, but from his (and some archaeologists') perspective, the Indo-Iranians who eventually found their way to the IVC were some sort of hybrid population between Sintashta-Petrovka-derived and BMAC cultures.

    The archaeological evidence just doesn't support Narasimhan's three-way model, particularly for SC/NW S Asians. I wouldn't rely on those. At the least, I'd include Namazga_EN (or Gonur1_BA). Even if it's just with G25. If you don't want to run my AASI simulations, then Namazga_EN + Sintashta + SiSBA2&3 + SSo should work (you could always infer the absolute AASI values from the latter two).

    This has been discussed before, but there's plenty of evidence of trade between Elam, the BMAC and the IVC, so there probably was some autosomal contribution from BMAC artisans to the IVC. Recall that SiSBA2 and 3 are probably migrants from somewhere near the Indus, so there's no reason to expect that migration in the opposite direction didn't happen.

    Those are two reasons why Narasimhan's three-way model was inappropriate.
    Last edited by DMXX; 08-08-2019 at 12:17 PM. Reason: comments; edits

  6. The Following 11 Users Say Thank You to DMXX For This Useful Post:

     bmoney (08-13-2019),  Kulin (08-08-2019),  laltota (08-08-2019),  MonkeyDLuffy (08-08-2019),  parasar (08-08-2019),  pegasus (08-08-2019),  Reza (08-08-2019),  Ruderico (08-08-2019),  Sapporo (08-08-2019),  scobar (08-09-2019),  Yggdrasi^ (08-08-2019)

  7. #14
    Registered Users
    Posts
    1,124
    Sex
    Location
    US/ India
    Ethnicity
    Punjabi Khatri/J&K/Multan
    Nationality
    India
    mtDNA (M)
    U2b
    Y-DNA (P)
    J2b2a

    India United States of America India Punjab Jammu and Kashmir
    Quote Originally Posted by Sapporo View Post
    The model itself is wrong as it ignores the BMAC contribution to South Asia. It's been discussed ad nauseam before but I've pointed out that there is likely a combination of mislabeling and/or sampling issues of the Jatt samples. This also occurred in the later Daamgard paper and we confirmed it with the few outliers among the Khatri and Kamboj samples, which were publicly available. The difference is that with Jatt Sikhs in Punjab is they are a much more numerous population (than either Khatris or Kamboj) and it is common for non Jatts/mixed individuals in Punjab to identify as such/use their surnames for perceived socio-cultural gain.

    If the Narasimhan samples become available, we can confirm it as well by running them through GEDMatch.
    This problem is starting to, and getting much worse in India. Go to Rajasthan and UP, and you'll find an enormous number of Rajputs. I have asked them about their lineage and they say they worship some old king/ priest. There are a few communities that lie about being Brahmins. Within Haryana Choudhary and Malik have gotten common among others. Go to Delhi and you see backward caste taking Khatri names. In Punjab it's Jatt. Within a generation or two the lie will get lost and they will start calling themselves these to even researchers. I know one family that I am almost certain lies about being a Jat.

    I am not blaming them. Backward caste status comes with a lot of disadvantages in India. It's just that then we start getting these odd results classified as diversity. I don't think diversity is a Kamboj (which is a North Indus population) coming out as a Chamar. I have looked through a few dozen Khatri/ Arora kits on Gedmatch and they all range between 6-14 NE Euro and 13-19 Caucasian, the lowest SI I have seen is about 25.
    Last edited by agent_lime; 08-08-2019 at 01:18 PM.

  8. The Following User Says Thank You to agent_lime For This Useful Post:

     Sapporo (08-08-2019)

  9. #15
    Global Moderator
    Posts
    645
    Sex
    Location
    UK
    Ethnicity
    Bengali
    mtDNA (M)
    M4b1
    Y-DNA (P)
    R1a-L657

    I take on board the point about accuracy and labelling samples. If that is true, it would certainly impact results and assumptions made from them. As sapporo says, if you could get hold of the academic samples, you could run them against other known samples and see what is mislabelling vs outliers.

    But much of what you describe above is probably part and parcel of the ethnogenesis of many communities? Why do we have such heterogeneity in y dna lineages, or spectrum of autosomal components if communities haven't married in / admixed / become associated with / "lied".

    People sometimes take on a very purist attitude on this forum, as if a blue blooded lineage exists for many communities that has had no mixing for the past 2000 years. And that this can be tested for autosomally. Even with Brahmins you see local mixing.

    There are too many unknown variables. We don't know the origins of rajputs, or how many waves of Brahmins moved into various areas (ie Bengal), or which castes mixed in the past and then stopped, or even when these castes were formed, or whether some Sikhs had mixed Muslim ancestors. Chowdhury, maliks are all semi recent terms. Landowners rise and fall. Over centuries, some will disappear from memory.

    We're seeing all the above through the prism of today's understanding.

    We may never know much of the history of these groups, but from a genomic point of view, we can at least get a better idea of what the intra group variability is.. The standard deviation within any group. So that we don't mislabel people as outliers, or missampled, or say they're lying etc etc.

    Whatever samples we may collect of gedmatch are always going to be skewed towards middle class, migrant population groups so the sampling bias will be high, when dealing with a population of over a billion. That is the potential benefit of 'some' academic papers if they are set up well.

    Until we ascertain a true understanding of the variability, then I would caution make firm judgement calls on who is making up what association with whichever group.

    My tuppence..
    Paternal; Y-DNA: R1a-L657> Y6> Y11> Y920*
    Paternal; mtDNA: M5b'c
    Maternal; Y-DNA: R1a-L657> Y6
    Maternal; mtDNA: M4b1

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

     bmoney (08-13-2019),  DMXX (08-08-2019),  FrostAssassin0701 (08-08-2019),  Kulin (08-08-2019),  laltota (08-08-2019),  Orchid (08-08-2019),  parasar (08-08-2019),  Rustyshakelford (08-14-2019),  scobar (08-09-2019),  Yggdrasi^ (08-08-2019)

  11. #16
    Administrator
    Posts
    3,688
    Sex
    mtDNA (M)
    D4j5*
    Y-DNA (P)
    R2a*-M124 (L295-)

    England
    Great point. And I've been thinking about this recently.

    It isn't clear to me whether the pop gen academics have a conventional statistical definition for what they consider an "outlier" from a main group. It also isn't clear that, should such a convention exist, whether it's applied to both ancient and modern groups without other preconditions or considerations. From what I've seen so far, that distinction is quantifiable, but it's based on an arbitrary parameter (f.ex. in Mathieson et al. with the Sintashta ones). Carlos from indo-european.eu has a similar perspective, so this doesn't look like a case of simple ignorance on my part.

    Another point to consider is that, even within groups/regions, intra-component variation exists. So, if one were to adopt a statistical convention based on a standardised assessment of a relatively homogeneous group (f.ex. the Cornish, Yazidi Kurds or Chechens), the application of that criteria will result in an unduly large number of "outliers" in populations who may naturally have a wider spread about the mean (or an inherent statistical shift in a particular direction).

    The above is relevant to the discussion because it highlights that our criteria for "typicality" in human pop genetics is both ambiguous and arbitrary, which will hamper attempts to screen out individuals with substantial cryptic ancestry from regionally proximal populations based on sociological considerations.

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

     agent_lime (08-08-2019),  laltota (08-08-2019),  Reza (08-08-2019)

  13. #17
    Gold Class Member
    Posts
    1,755
    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 Reza View Post
    I take on board the point about accuracy and labelling samples. If that is true, it would certainly impact results and assumptions made from them. As sapporo says, if you could get hold of the academic samples, you could run them against other known samples and see what is mislabelling vs outliers.

    But much of what you describe above is probably part and parcel of the ethnogenesis of many communities? Why do we have such heterogeneity in y dna lineages, or spectrum of autosomal components if communities haven't married in / admixed / become associated with / "lied".

    People sometimes take on a very purist attitude on this forum, as if a blue blooded lineage exists for many communities that has had no mixing for the past 2000 years. And that this can be tested for autosomally. Even with Brahmins you see local mixing.
    I can't speak for other ethnic groups in South Asia but the Jatt Y-DNA study by Mahal (with a sample size of around 200 individuals from India I believe) shows that Jatt Sikhs and even Hindu Jaats are overwhelmingly R1a1a and L1a (mostly L1a2 with a minority of L1a1). My own haplogroup spreadsheet data (collected via 23andMe or FTDNA primarily) agrees with the study itself (the vast majority are L1a or R1a1a-Y7). The rest are mostly strong minorities of R2, J, Q and H.

    It isn't necessarily about "purity." There were clearly multiple admixture events across South Asian that formed the ethno-religious groups we see today. However, many of them have been relatively endgamous more recently. Otherwise, you wouldn't see Brahmins across India retaining Steppe admixture or stark contrasts in autosomal ancestry between neighboring populations such as Hindu Jaats/Ror and neighboring Chamar populations. Nor would be able to see the strong clusters seen in PCA modeling or genetic tools such as nMonte.

    I have over 100 Jatt Sikh GEDMatch kits. There is a range in their primary admixture components but it's quite a small standard deviation for a group numbering in the millions. If individuals are scoring multiple standard deviations (2.5+ as an example) from the average, I find it difficult to just label them outliers that are party of the diversity rather than mislabeled or recently mixed.

    I'm always open to evidence proving the contrary though. If you can find 30-40 Jatt Sikh kits or individuals scoring 10% more AASI than Khatris, I'm all ears. The Narasimhan study has 41 samples and that is their average. Not individual outliers but the actual overall average. I can't even begin to fathom how diverse the samples potentially score to get such a large difference from Khatris.


    So that we don't mislabel people as outliers, or missampled, or say they're lying etc etc.

    Whatever samples we may collect of gedmatch are always going to be skewed towards middle class, migrant population groups so the sampling bias will be high, when dealing with a population of over a billion. That is the potential benefit of 'some' academic papers if they are set up well.
    You're free to believe they're potentially representative samples. Based on currently available data, I will continue to doubt it.

    Anyways, I should clarify that Jatt Sikhs are the dominant and most populous socio-cultural sub-ethnic group in Indian Punjab. Also, one of its most endogamous or "casteist" for a better lack of word. I'm not sure how familiar you are with Jatt Sikhs but I can assure you the migrants we have in the West aren't primarily descended from wealthy or mobile middle class backgrounds in Indian Punjab. Many of these migrants came over as peasant farmers 30+ years ago (including both sides of my family) without a penny to their name and had to initially leave their families behind while trying to make a better life for themselves and eventually their immediate family/cousins. I still have distant relatives back in Punjab (some are well off and others aren't). So, I question any potential sampling bias related to "class." I'm perplexed how those from a higher vs. lower socioeconomic class would even differ for such a specific sub-group.

    My primary point is you don't just uproot yourself from a livable/satisfactory middle class background to start from the bottom in a completely foreign country. It's far too risky.


    Until we ascertain a true understanding of the variability, then I would caution make firm judgement calls on who is making up what association with whichever group.

    My tuppence..
    A simple question. If we were to get a hold of the Jatt Sikh samples from the Narasimhan study, would you still argue that potential individuals scoring like Punjabi Chamars or other non Jatt North Indian populations, were representative/just outliers that are party of the diversity? If so, based on what? Heterogeneity of Y-DNA? The Mahal study already shows there are a few dominant lineages. Autosomal heterogeneity? How many GEDMatch kits or Global 25 coordinates would we need to find this autosomal heterogeneity?

    Also, I should note that all 41 of Narasimhan samples are supposed to be from Bathinda while my 100 kits are from diverse origins across Indian Punjab and some with ancestry from Pakistani Punjab. Yet, my data is what you are suggesting has a sampling bias? Finally, why do the Khatri and Lohana samples in the Narasimhan study match up with our GEDMatch data but the Jatt Sikh samples don't? Is it just a numbers game with 3 samples used for each of Khatris and Lohana vs. 41 for Jatt Sikhs? Even the single Jaat sample from the Pathak paper matches up with the Hindu Jaat samples we have from GEDMatch and to prashant's cousin.
    Last edited by Sapporo; 08-09-2019 at 07:24 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": [

  14. The Following 7 Users Say Thank You to Sapporo For This Useful Post:

     agent_lime (08-08-2019),  bol_nat (08-08-2019),  Kulin (08-08-2019),  laltota (08-08-2019),  MonkeyDLuffy (08-08-2019),  pegasus (08-08-2019),  prashantvaidwan (08-08-2019)

  15. #18
    Registered Users
    Posts
    1,124
    Sex
    Location
    US/ India
    Ethnicity
    Punjabi Khatri/J&K/Multan
    Nationality
    India
    mtDNA (M)
    U2b
    Y-DNA (P)
    J2b2a

    India United States of America India Punjab Jammu and Kashmir
    Quote Originally Posted by Reza View Post
    I take on board the point about accuracy and labelling samples. If that is true, it would certainly impact results and assumptions made from them. As sapporo says, if you could get hold of the academic samples, you could run them against other known samples and see what is mislabelling vs outliers.

    But much of what you describe above is probably part and parcel of the ethnogenesis of many communities? Why do we have such heterogeneity in y dna lineages, or spectrum of autosomal components if communities haven't married in / admixed / become associated with / "lied".

    People sometimes take on a very purist attitude on this forum, as if a blue blooded lineage exists for many communities that has had no mixing for the past 2000 years. And that this can be tested for autosomally. Even with Brahmins you see local mixing.

    There are too many unknown variables. We don't know the origins of rajputs, or how many waves of Brahmins moved into various areas (ie Bengal), or which castes mixed in the past and then stopped, or even when these castes were formed, or whether some Sikhs had mixed Muslim ancestors. Chowdhury, maliks are all semi recent terms. Landowners rise and fall. Over centuries, some will disappear from memory.

    We're seeing all the above through the prism of today's understanding.

    We may never know much of the history of these groups, but from a genomic point of view, we can at least get a better idea of what the intra group variability is.. The standard deviation within any group. So that we don't mislabel people as outliers, or missampled, or say they're lying etc etc.

    Whatever samples we may collect of gedmatch are always going to be skewed towards middle class, migrant population groups so the sampling bias will be high, when dealing with a population of over a billion. That is the potential benefit of 'some' academic papers if they are set up well.

    Until we ascertain a true understanding of the variability, then I would caution make firm judgement calls on who is making up what association with whichever group.

    My tuppence..
    What you are describing is probably how many castes came to be. True. And of course, I don't think there is anything such as purity of caste. People will always mix in. People want to be associated with the higher rung of society. Rajputs originally came from "Raja Putra", it was originally a title given to direct descendants of Royal lineage. Today a lot of folks will have a "Kul Devta"(family diety) and then change the last name when their children are born to Singh or Rajput.

    This happened on a large scale in the last few dozen years as well. People would change their last name to Kumar, or Valmiki. There are other factors too, people change last names when moving to bigger towns. They have the right to change their name, no one can stop them. As for academic samples the issue becomes pronounced when people themselves don't know the truth. There will be an explosion of such samples coming from Northern India in the next 10-20 years. There are also communities that accept marriage outside, for example Tamil Brahmins were trying to import Brahmin brides from Uttar Pradesh. Many city dwelling folks have mixed caste marriages, depending on their communities they will be accepted, some communities ostracize them. If I marry a Rajput, even though society doesn't know better, should my kids be looked as academic samples for Khatris? My paternal family went through a similar thing when an uncommon last name forced my granddad (who was probably sick of people asking what caste we are) to change his kids last name to some more common.

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

     MonkeyDLuffy (08-08-2019),  Sapporo (08-08-2019)

  17. #19
    Global Moderator
    Posts
    645
    Sex
    Location
    UK
    Ethnicity
    Bengali
    mtDNA (M)
    M4b1
    Y-DNA (P)
    R1a-L657

    Sapporo, I wasn't actually replying to you regarding Jatts, it was actually to agent_lime with this specific paragraph which I ought to have quoted:

    This problem is starting to, and getting much worse in India. Go to Rajasthan and UP, and you'll find an enormous number of Rajputs. I have asked them about their lineage and they say they worship some old king/ priest. There are a few communities that lie about being Brahmins. Within Haryana Choudhary and Malik have gotten common among others. Go to Delhi and you see backward caste taking Khatri names. In Punjab it's Jatt. Within a generation or two the lie will get lost and they will start calling themselves these to even researchers. I know one family that I am almost certain lies about being a Jat.
    So I wasn't really commenting on jatt sikh samples from the academic papers nor suggesting they were representative. I actually acknowledged your point about specific mislabelling and agree that you could confirm your theory if only you could get hold of the samples.

    Anyhow, my point still stands as a general principle that in the West, the migrant populations are skewed towards middle class backgrounds. I didn't say wealthy or upper middle class. But generally speaking, migrants don't come from the bottom of society ie the scheduled castes, the labourers or tribal populations. And in my experience of Punjabis in the UK from both Pakistan and India, I think that holds. People migrate for a number of reasons. And economic opportunities are one of them.

    Gedmatch samples are great in terms of numbers, but will be full of selection bias. Relatives of relatives as well as socioeconomic implications (you have to be well off enough to undertake a commercial dna test). They cannot be extrapolated to represent a population of millions unless you apply vigorous filters.

    I have no vested interests in this topic given my background. But if definitive conclusions are to made, then I think using the approach dmxx suggested regarding intra group variability, as well as population representation and sampling bias, should be taken into consideration.
    Paternal; Y-DNA: R1a-L657> Y6> Y11> Y920*
    Paternal; mtDNA: M5b'c
    Maternal; Y-DNA: R1a-L657> Y6
    Maternal; mtDNA: M4b1

  18. The Following 4 Users Say Thank You to Reza For This Useful Post:

     bmoney (08-13-2019),  poi (08-08-2019),  scobar (08-09-2019),  Yggdrasi^ (08-08-2019)

  19. #20
    Administrator
    Posts
    3,688
    Sex
    mtDNA (M)
    D4j5*
    Y-DNA (P)
    R2a*-M124 (L295-)

    England
    Reza's correct. I don't understand the basis of your objection (Sapporo).

    Higher socioeconomic classes are heavily over-represented in consumer genomics testing, and there are all sorts of implicit selection biases when it comes to recruiting participants for studies in the field (non-medical human studies seem to be worse in controlling for these variables).

    It looks like you've also misread Reza regarding the extent of AASI in Punjabi Jatts - His contention is that the current average should not be taken as representative with absolute certainty. Qualifying an outcome as being less secure isn't the same as arguing that reality is contrary to said outcomes. I'm also unsure if he was even discussing them.

    Using uniparentals to justify a particular autosomal trend is also a very weak argument. We're all aware of the Y-DNA to mtDNA to autosome asymmetry in the Subcontinent. Per that reasoning, we'd expect the Kyrgyz and certain northern Indian populations to be massively MLBA steppe-derived given the predominance of Y-DNA R1a1a among them. Isn't so. Founder effect.

    Regarding how "representative" the current roster of Punjabi Jatts are when it comes to AASI - A type 2 error calculation will tell you that.

  20. The Following 6 Users Say Thank You to DMXX For This Useful Post:

     bmoney (08-13-2019),  FrostAssassin0701 (08-08-2019),  laltota (08-08-2019),  poi (08-08-2019),  scobar (08-09-2019),  Yggdrasi^ (08-08-2019)

Page 2 of 14 FirstFirst 123412 ... LastLast

Similar Threads

  1. Cladogram from STR Values
    By jbarry6899 in forum Other
    Replies: 16
    Last Post: 12-08-2016, 06:23 PM
  2. Page Help for Y-DNA - Standard Y-STR Values
    By Mac von Frankfurt in forum FTDNA
    Replies: 2
    Last Post: 11-30-2015, 02:55 AM
  3. Y-DNA NGS testing and STR values
    By Salkin in forum General
    Replies: 0
    Last Post: 01-17-2015, 07:54 PM
  4. R1b-L21 STR Allele Values and Frequencies
    By MikeWhalen in forum R1b General
    Replies: 0
    Last Post: 11-22-2014, 05:45 PM
  5. Different Values for DYS487
    By Caesarea in forum FTDNA
    Replies: 0
    Last Post: 04-18-2014, 10:49 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
  •