Admixture (Ref3 K=11) HRP0141-HRP0150

Here are the admixture results using Reference 3 for Harappa participants HRP0141 to HRP0150.

You can see the participant results in a spreadsheet as well as their ethnic breakdowns and the reference population results.

Here's our bar chart and table. Remember you can click on the legend or the table headers to sort.

If the above interactive charts are not working, here's a static bar graph.


  1. Zack

    It would be great, if instead of HRPxyz, u (also) mention the ethnicity on the graph. Also, will be great if regional ethnicities could be grouped together. More work for you, but just wondering if its easy for you.


    • Putting the ethnicities on the bar chart would add too much clutter.

      You can sort the ethnicity column in the table below the chart and that groups those from same ethnicity together in the bar chart.

      Now that we have a number of ethnicities with multiple participants, I'll start posting average Admixture results too.

  2. first full romany snp autosome i've seen. would like to know more about this romany.

  3. Interesting that the part Serb-part Roma participant has more Onge admixture than the newest Roma (assuming he/she is fully Gypsy).

    • there's a big difference between east european roma and other gypsy groups. finnish gypsies don't look at indian at all. they're brunette though.

  4. I have calculated ANI, ASI, EA and AFR components for the Ref. 3 populations and for all HRP participants without using the Onge fraction. I added the SW Asian and European as well as a portion of the South Asian components which I adjusted to fit the Reich published values of ANI. You can see the spreadsheet below. For HRP0001 to HRP0090, I repeated using Reference 1 data.

    • I forgot to add that the ANI and ASI results will be meaningless for those without much South Asian ancestry with negative numbers and numbers over 100%!

    • Other than the negative or larger than 100% values, the numbers for some (e.g., mixed and Romani participants) look wrong.

      Did you use linear regression? How was the fit?

      • Zack, is it right to separate out the 1-4% East Eurasian admixture that non-Bengali/Chota Nagpur South-Asians get from the actual ASI component, like Balaji has done? East Eurasian (i.e Siberian, Papuan, American, etc) seems rather noisy in Indians, and thus I was under the impression those minor components popping are simply an artifact of ASI's marginal East Eurasian affinity.

      • I calculated ANI as SW Asian + European + 32% + 0.18*S Asian. The difference between the calculated values and the Reich values was 3.6% in the worst case. Of course this does not work for those with low South Asian values - even a person with 0% South Asian will have 32% added to his ANI!

        But this method does work for people who fall in the middle of the Indian Cline. For example, with HRP0016, I calculate ANI of 60%, you calculate 61% and Dienekes 62%.

        With the part-Romani individual HRP0015, I do not get any meaningful value for ASI using the Ref. 3 K=11 data. However, I can do a little better with the Ref. 1 K=12 data. Here, I calculate ANI as Baloch + Kalash + SW Asian + European + 4% + 0.4* S Asian. With this, I get 5% ASI for HRP0015. I think this is an underestimate. Using your method, we get 14% ASI which I think is an overestimate. So let ua put his ASI as 10%. Assuming his Indian ancestors had equal parts ASI and ANI, we can say he is about 20% Indian!

        HRP00159 is lower in South Asian ancestry. Let us put him as 10% Indian!

        • I'm still at a great loss with my results. It shows no Onge at all - I have a Gujarati g-grandmother EVERY grandparent I have is no less than 1/2 Romany (the same type as HRP0149). In fact, that individual has the same last name as my paternal grandmother (that means they're are the from same as my grandmother, even though we've not yet made the paper trail connection.

          Also, I and every other individual with my last name tested, have resulted in a Y-DNA of R1b* - that too is surprising but it simply shows a common ancestor who must have been fathered by a non-gypsy.

          I'm no expert in the field of DNA, however even Dr. McDonald found my results to show 3.67% from India. Can anyone explain? Your assistance would be greatly appreciated.

          • The Indian in Dr. McDonald's results is a combination of Ancestral North Indian and Ancestral South Indian. Onge component here is correlated with Ancestral South Indian only. However, also remember that the Onge component underestimates the ASI ancestry.

            So it is possible that you might have a small amount of ASI which didn't show up in my analysis. It is also possible that your South Asian ancestors were from the northwest of the region and thus a lot more ANI than ASI. Add in the admixture over the centuries and you might not have inherited much ASI SNPs.

            I am working on something to estimate ASI percentage better than the current system with Reference 3. Let's see what results you get then.

          • Zack - Thank you so much for your response and your explanation. Also, as I've already explained, my g-grandmother was Gujarati, however she was Parsi. Would that also play a role in the low percentage of ASI?

          • Her being Parsi would definitely result in a lower ASI percentage. I am not familiar with any autosomal genome ancestry analysis of the Parsi community, but the Y-DNA studies showed them closer to Iranians and an mtDNA study concluded they were close to Gujaratis. So Parsis are likely a mix of Indian and Iranian, thus having low ASI percentages.

          • Also can I change your ethnicity info in the spreadsheet by replacing Gujarati with Parsi?

          • For reference, Parsi mtDNA demographics:
            M* - 54.5%
            U4 - 13.6%
            HV2 - 9.1%
            T1 - 6.8%
            T* - 4.5%
            U1 - 4.5%
            HV* - 2.3%
            H - 2.3%
            U7 - 2.3%

            Also, they seem to exhibit more R2 than generic Iranian populations, which might indicate assimilation of Western Indian males into their fold.

          • Parsi y-DNA frequencies (Gujarati Parsees; Qamar et al. 2002):
            24/90 = 26.7% P(xR1a)
            3/90 = 3.3% Y*(xA, C, DE, H2, J, K)
            7/90 = 7.8% R1a
            35/90 = 38.9% J
            5/90 = 5.6% E(xE1b1a)
            16/90 = 17.8% L

          • Yes, absolutely - if it fits your studies better. I didn't say anything because she is a Patel (I'm sorry I don't know the whole background to respond as to why she's a Patel and a Parsi)and she spoke only Gujarati. She died when I was 7.

          • A Parsi and a Patel? I am confused, but then my knowledge of South Asian ethnicities is very superficial. May be one of the other participants can shed some light.

          • You may also want to change Romany to Romany Gypsy from England so it matches HRP149.

          • No, no - please allow me to great grandmother was a Parsi. That I know. I also know that her father (a Patel) was hindu and her mother Parsi. I can imagine a scandal but never learned the story because anyone who could have answered the question as I got older and became interested...died. I can't imagine what the family was like but all said and done that Parsi Patel married a Gypsy who was in India with the English army. I sure wish I had these answers. I've always wanted to know.

          • ...and an Edit - My G-Grandmother did not die when I was Grandmother (her daughter) died when I was 7. My apologies for misleading.

          • Zack,
            There are a number of Parsis in Gujarat with a Patel title. Their first names are usually Iranian.

  5. I am sad to find that there is no siberian nor east asian admixture in my case(HRP142), considering my mtDNA Hap is c4a1. Am I the only C4a1 South Indian around??
    Thanks for the results Zack!

    • Also, unless I am mistaken, I think Tamil Brahmin and Tamil Brahmin Iyer all refer to the same group of people. I don't know how much trouble it would be, but can we remove the 'Iyer' suffix from these to simplify? Maybe someone can correct me if I am wrong on this.

      • Also, unless I am mistaken, I think Tamil Brahmin and Tamil Brahmin Iyer all refer to the same group of people. I don't know how much trouble it would be, but can we remove the 'Iyer' suffix from these to simplify? Maybe someone can correct me if I am wrong on this.

        iyengar are iyers? (i don't know anything about this)

        • Pretty much. The split happened around the 10th century A.D in south India, over the definition of God. In very simplistic terms, the Iyers 'typically' follow "non-dualism"(advaita vedanta), whereas the Iyengars have a more personal view of God.

          • Apart from the theological differences, some believe that Ramanujam converted low-caste people to Iyengars and that therefore Iyengars are not "true" Brahmins! However the Harappa data do not seem to support this!

          • I would've had great respect for him if he had done so! He would then be the first "Hindu" in recorded history who tried to bring equality and reform.

    • Prior to the Iyer-Iyengar division among Tamil-speaking Brahmins, it is safe to assume Vaishnavaite and Shaivaite Brahmins intermarried with each other. In fact, some sources allude that the theological preference (Shavaite vs Vaishnavaite) was largely specific to a family and subject to individual choice prior to the advent of Sri Ramanujacharya. So yes, Iyengars as an endogamous group is a recent phenomenon.

      As for the rumors that Sri Ramanuja converted lower castes to caste Brahmin status, the reality of it must be between the two extreme notions (no conversions vs large-scale conversions). To my knowledge Ramanuja simply preached Visishta Advaiata to the lower castes, and no conversions took place per se. However, the term Sri Vaishnava today refers exclusively to followers of Ramanuja's Visishta Advaita, namely Iyengar Brahmins. It is specifically rumored that the Thenkalai subsect of Iyengar Brahmins might partially descend from recent non-Brahmin converts. However, I think such notions have sprung up due to the theological differences the Vadakalai and Thenkalai sub-sects share. The minor differences between Thenkalais and Vadagalais are mainly doctrinal, and sometimes rather petty - for example, the Vadagalais sport a U-shaped sricharnam, whereas the Thengalais a V-shaped sricharnam (religious insignia). The Vadaglais are noted to ascribe greater importance to the Vedas which are in Sanskrit (and consequently also Vedic rituals), whereas the Thengalais stress the importance of the Divya Prabandams of the Alwars which were in Tamil, as it offered a message of devotion which cut across caste, ethnic and gender lines. In light of the same, it is believed that the Thenkalais as a group were partially formed by the absorption of some non-Brahmin Vaishnavas. This distinction in itself is a recent phenomenon, and in reality both sects have integrated both beliefs (i.e emphasis on the Vedic rituals and philosophy, coupled with the philosophy of the Alwars) into their theological beliefs.

      So far, there are two Tamil Nadu Iyengars, one is HRP72 and the other HRP48. Going by a correspondence I had with the latter a while ago on 23andMe, he is a Vadakalai, as is HRP72. While I predict that we'll see no real difference between Vadakalais and Thenkalais, it'd be worth it to see whether Thenkalais will differ from the Thenkalais. The other two Iyengars from Karnataka are also assumably belong to the Vadakalai sub-sect, as the vast majority seem to be, with a small Thenkalai minority.

      It's also worth noting that in my experience, within the Iyengar sub-sect itself, there is a (silly) stereotype/notion that is widespread that the Vadakalai Iyengars are fairer skinned and more North-Western looking as opposed to the Thengalais who tend to be darker and more non-Brahmin looking. Of course both groups run the gamut of phenotypes.

      Regarding mtDNA C4a1, I proposed in the previous ADMIXTURE batch's comments that it might be fun to make a spreadsheet with the participants' haplogroups.

      "If people are that interested in the y-DNA/mtDNA haplogroup demographics across India, we could probably request Zack to make a public spreadsheet, editable via a Google account wherein participants (or participants via a message to Zack?) could enter their paternal and maternal haplogroups against their alphanumeric ID. This, perhaps combined with the y-DNA results from the India DNA project over at FTDNA should shed some light regarding the y-DNA and mtDNA distribution among South-Asia's ethnic groups. This should be useful especially for mtDNA as it hasn't been studied on the basis of region or ethnicity in South-Asia (in any peer-reviewed study to date). y-DNA on the other hand seems fairly predictable."

  6. Zack, when I click on the "Harappa Participants Haplogroups", it links back to Balaji's ANI/ASI spreadsheet; not SB's haplogroup spreadsheet.

  7. Harappa Participants Haplogroups | Harappa Ancestry Project - pingback on July 22, 2011 at 12:23 pm
  8. From looks i would say im much more than 20% indian only, my family from moms side looks very indian and even my dad is a 1/4 roma. I doubt im only 20% indian. sorry guys

Trackbacks and Pingbacks: