Improved Admixture Bar Charts

I have improved the Admixture bar charts further. As per your demands, ethnicity information is now available in a table right below the bar plot, in the same order as the bar plot IDs.

Also, you can click on any of the legend color rectangles on the right to sort the bar chart and the table by that ancestral component. Similarly, click on the header row of the table to sort by a column.

I might make some minor tweaks to this one.


  1. Great format. It looks good. The only thing that I wonder would be if you could, on the main chart, put the ethnic information in parantheses after the identification number? That way, all information would be available at a glance. Though, maybe you tried that already, and it doesn't work or doesn't fit your goals.

    BTW, I noticed something interesting. The Gujurati reference population, at K=2, looks rather similar in South Asian, Pakistani/Caucasian, and Kalash components, to HRP0009. That seems rather strange to me, considering the geographic differences involved. However, I wonder if this has something to do with the split in the Gujurati group that Razib has discussed in some of his blog posts?

    • Sorry, that should have said "K=12" rather than "K=2".

    • Because I want these charts to be less than 600 pixels wide, putting the ethnic info (which is quite long) makes the bar plot looking odd.

      The Gujarati average ancestry is not a good data point to compare against because of the two clusters that Razib blogged about.

      I am thinking of putting individual reference dataset results online too. Just need to figure out a way to present it.

  2. However, I wonder if this has something to do with the split in the Gujurati group that Razib has discussed in some of his blog posts?

    "gujarati_B" is *very* south asian. the two groups get very distinctive K > 10

    • Interesting. Thank you for the link. I can see the importance of determining who (which jati) Gujarati_B actually is.

      Do you know who the Sindhis are? Can they be considered representative of Sindh as a whole? To me, one of the most striking results is how different the Sindhi and Gujarati groups are, despite being neighboring regions.

      • There are a few Sindhis with African admixture in the dataset.

        Also my understanding is that in terms of language, history and genetics, Sindhis are much closer to Punjabis than to Gujaratis.

        • Yes that is very true Cholistan, Thar and Rann of Kutch are stronger barriers than between Sindh & Western Punjab. After all there are x million Seraiki speakers who are intermediate between the two (and have the syncretic Sufi culture we all love) however Kutchi is only spoken by a few Hindu/Muslim merchant castes and in Kutch.

          Mind you though that in Punjab (incl Haryana), Sindh, Rajasthan and Gujarat there are so many overlapping tribes, clans and castes that there is a fair degree of intermixture.

        • The sindhis with african admixture are most likely siddis or at least part siddi.

  3. { Brown Pundits } » Harappa Ancestry Project update - pingback on February 24, 2011 at 2:23 pm
  4. The Sheedis of Sind (or Gola/Gadra as they are known in Baluchistan) are mostly 'pure' African with little admixture.

    That said, there are probably quite a few Balochis/Sindhis/Brahuis and specially Makranis (Southern Baluchis from the coastal areas) that have absorbed some African genes from them.

    • I don't think they are pure African, though they have a significant African admixture.

      • the probability that they're pure african is very low if they've been around that long, unless indians are the most racist people in the world. more racist than any other people who have existed šŸ™‚ also, i've seen photo essays of these people. they seem to exhibit indian admixture in their features to me quite often.