Wednesday, September 14, 2011

'africa9' calculator

I have devised a new calculator targeted specifically for Africans. Admixture proportions in the reference panel, Fst distances between the K=9 components, as well as individual results for Project participants from the North_Africa_D, North_African_Jews_D and East_African_D populations can be seen in the spreadsheet.

The calculator combines data from Henn et al. (2011), HGDP, and Behar et al. (2010). As a result, the number of SNPs is small: there is probably noise in the minor components, but the major components of one's ancestry should be well-defined.

It should be used only by Africans and African-West Eurasian admixed individuals. It is not meant for people with additional admixture (e.g., South/East Asian or Native American).

You can download the calculator RAR from here (Google docs), or here (sendspace).

You need to extract the contents of the RAR file to the working directory of DIYDodecad. You use it by following exactly the instructions of the DIYDodecad README, but always type 'africa9' instead of 'dv3' in these instructions.

Terms of use: 'africa9', including all files in the downloaded RAR file is free for non-commercial personal use. Commercial uses are forbidden. Contact me for non-personal uses of the calculator.

NB: Note that the components of 'africa9' do not necessarily have the same meaning as the same-named components you might have seen elsewhere. Refer to the spreadsheet for the admixture proportions and Fst distances between components. For example, the NW African is substantially removed from other West Eurasian components in Dodecad v3 but equidistant from Europe and SW_Asia in 'africa9'. I also advise that you read Interpretation of ADMIXTURE results: component sharing


  1. Any help with this error I have?

    Chromosome#22 position: 15467656 to 0
    window size: 500 SNPs | -15.47Mb

    Traceback: not available, compile with -ftrace=frame or -ftrace=full
    Fortran runtime error: Array section out of bounds
    Warning message:
    running command 'DIYDodecadWin africa9.par' had status 2

  2. I'm guessing there may not be even 500 SNPs on Chromosome #22, since 'africa9' is based on a smaller number of SNPs.

    Do a "bychr" to see how many SNPs there are on #22, and then make sure the window is no greater than that number. Actually, if you want to have window sizes of equivalent physical length to 'dv3', you should have a window scaled by the total number of SNPs, or about 80 SNPs.

  3. Since your project is the only one giving the opportunity to investigate a litte the African diversity, I'd like to ask you something: is it possible to get rid of the Northwest African component and eventually to have a middle east component that do not contains too many East African alleles?
    I tell you this because the results are not very satisfying for complex populations such as Berbers. A question we would like to solve is where our Africanity comes from. The northwest African component is again hiding things (both Eurasian and African) and I find it weird to have heterogenous results between Tunisians, Algerians and North Moroccans (the other related groups too but they contain recently SSA admixed individuals).
    Mozabites are again probably skewing the results and it also affects the West Eurasian part.

    Also, East Africans (Horners) results are a bit off (a bit like 23andme) because they imply they are mainly or half West Eurasian-derived when in fact they are mainly African with substantial West Eurasian ancestry.

  4. Also, East Africans (Horners) results are a bit off (a bit like 23andme) because they imply they are mainly or half West Eurasian-derived when in fact they are mainly African with substantial West Eurasian ancestry.

    Both North and East Africans appear to be partly of West Eurasian origin. They invariably end up stretched along the line linking Sub-Saharan Africans with West Eurasians, which is a good sign that they have this type of dual admixture.

    It appears that the West Eurasian component is different in North Africans and in East Africans; in the former case it is more "Mediterranean", in the latter it is more "Southwest Asian". These components have intermixed with native Africans long time ago, so today there are stabilized blends that are the Mozabite-centered and East African-centered components.

    There is also the issue of ascertainment bias. Current microarray chips are not particularly optimized for most Africans who are very diverse genetically. So, while we can definitely discern the existence of many African clusters, their relationships are less clear.

    Finally, I'll keep your request in mind. It is probably a good idea to do a "global" calculator with only the major components (i.e., no East African or Northwest African ones); you can kinda look how this will be here:

  5. Dear team,

    I am looking for the correct definition of the fields below. Please tell me what they mean:

    1.) 1st Population Bamoun, Kaba, Kongo, etc. - what does this mean?
    2.) What does the figures represent? Bamoun @9.279 - What is this calculation

    These results are from the,
    Click on Admixture-proportions analysis
    Select: Dodecad and click on Admixture Proportions (with link to Oracle) and click continue
    Enter your kit number and select Africa9 calculator model. Then click continue.
    Once your results appear, click Oracle-4
    Copy and past the “Using 1 population approximation:” (as shown below)

    Thank you!

  6. Hello,
    First I'll say thanks for your time and your hard work completing this for people to use! At first I didn't see that this was for African and or West Eurasian admixed individuals. I recently through this site found out that I have an African ancestor is why I used this. I once took the Eurogenes K9b test and I showed almost 2% African. Almost one percent was from South Africa. Which I later learned is where the San tribe is from. And again on this test I show 0.97% San. With 0.25% Mbuti. I wish there were another calculator that utilized the San population.

  7. I am southeast Asian with a little sprinkling of European blood (during Spanish colonial times in the Philippines, sometime early 1800s). I had the Africa9 admixture calculator, and the result is shown below:

    Europe 40.55%
    NW_Africa -
    SW_Asia 33.40%
    E_Africa 10.23%
    S_Africa 4.39%
    Mbuti 0.85%
    W_Africa 7.92%
    Biaka 2.00%
    San 0.65%

    What is the significance of two high percentage results (Europe 40.55% and SW Asia 33.40%) considering that my y-dna Hg (O-M175) is southeast Asian?

  8. Will the Oracle calculators be brought back on Africa9 Admixture Proportions? None of the Oracle X calculators work on Gedmatch.

  9. Hi! I was wondering.. Will there also be Europeans' results up in the spreadsheet? I'm mostly Dutch, with some minor Spanish. I wanted to see if there's any so called "Moorish blood" running through my Spanish roots. I only got 67.82% European and 5.95% NW African. The latter seems quite high actually, as on the other calculators it was only 0.03%/0.04%. I'd love to compare my results to other Europeans' results, like Dutch, British, French, Spanish, Portuguese..

  10. Hello. How feasible is this for African Americans? There were my results:

    11.45% European
    9.56 SW Asian
    4.76% East African
    10.69% South African
    1.69% Mbuti
    54.63% West African
    7.18% Biaka

    On 23AndMe, my Speculative results came back at 20% European, 78% African (5% Central and South African; 4% Broadly African) and 0.6% Native American/ East Asian and very little North African/ Middle Eastern to speak of. This seems drastically different.