Xing Ref3 K=11 Admixture

Xing et al dataset is interesting because it has a number of South Asian populations:

  • 25 Andhra Pradesh Brahmin
  • 10 Andhra Pradesh Madiga
  • 11 Andhra Pradesh Mala
  • 22 Irula
  • 25 Nepalese
  • 25 Punjabi Arain
  • 14 Tamil Nadu Brahmin
  • 12 Tamil Nadu Dalit

Unfortunately, the dataset does not have a lot of common SNPs with 23andme, FTDNA and the other data I am using.

However, I did run a reference 3 admixture on Xing data using about 30,000 SNPs. Since this is a lot less than the usual 118,000 SNPs, the noise levels are much larger.

Here is the spreadsheet with the Xing group averages for reference 3 admixture at K=11 ancestral components.


  1. Zack, this is amazing for me, since my maternal great grandfather was Tongan. This does indicate that the more East one moves in the Polynesian islands the lesser the Papuan influence.

    I am assuming these were only autosomal tests, and there wasn't any mtDNA or YDNA tested. If there were, is there a way to gain access to the data?


    • That's interesting about the Tongans and Samoans. I did not expect them to be so high in the East Asian component. So you are 1/8th Tongan?

      Xing et al did only autosomal genotyping. mtDNA or Y-DNA was not tested.

      • Well really 1/16, my great grandfather's father was a Tongan who invaded Fiji with a chief called Enele Ma'afu in the 1840's. They invaded in the guise of spreading Christianity, but instead were conquerors. They subdued most of the Eastern Islands, and pockets of the main island of Viti Levu.

  2. Thank you, Zack! What's interesting here is that the Irula and Tamil Dalits are entirely made up of the South Asian and Onge components.

    • The isolated populations such as the Irula seem to have much higher diversity that the more admixed ones such as Brahmans.

      "Although the HapMap GIH and the Brahmin populations have genetic diversity values comparable to those of other HapMap Eurasian populations, diversity values (Ï€ and H) in the Irula, Mala/Madiga, and Yadava samples are higher than those of the HapMap African populations."

  3. Could you also upload a spreadsheet with individual results for the reference populations? The Nepalese in specific have a lot of intra-ethnic variation, it seems. The Eurasian PCA plot in the actual study illustrated a cline for the Nepalese that spanned from the Punjabi Arains, while many other individuals deviated towards East-Eurasia.

  4. Hi, I recently run ADMIXTURE and MDS on a subset of the Global Dataset that you made publicly available about one year ago.
    Here is my entry on it:

    A question that I had was regarding the 8 Brong samples from Ghana that you had included in that Dataset, apparently, you specify Henn (2011) as your source in the .csv sheet, however, the supplemental data source for Henn(2011) does not specify a listing for this specific population, is this a typo or a mis-classification of the samples on your part, or were these samples included in the .bed files supplied by Henn(2011) but not published in the paper ?
    Thank you for the Dataset and the Very best of regards.

  5. Xing Ref3 Admixture South Asians | Harappa Ancestry Project - pingback on March 10, 2012 at 6:41 am
  6. Somone need to start explaining how these Onge admix keep appearing in some results keep but disappears in other results?

    I know Thai have about 9% Negrito mtDNA from Haplogroup M21a and M21b and 11%.I'm guessing that's what they mean by Onge? and if many Vietnamese intermarried with Thai, than their Onge admixture will break down to 4.5%?

    Malaysia also have 7% negrito mtDNA, 3% Papuan mtDNA and 3% Papuan Y-DNA.

Trackbacks and Pingbacks: