Explainer

ADMIXTURE

Do note that the admixture components do not necessarily represent real ancestral populations. Also, the names I have chosen for the components should be thought of as mnemonics to ease discussion. I chose them based on which modern populations in my data these components peaked in. They do not tell anything directly about ancestral/ancient populations. The best way to look at these admixture results is by comparing individuals and populations.

Leave a comment ?

13 Comments.

  1. Hi Zack

    Apologies if it's somewhere and I missed it, but how do you create the interactive barcharts for structure/admixture results?

  2. Hi Zack,

    I was just overlooking my Admixture results that you posted (tanks, btw!), and i'm wondering about the American classification. Is there something that can be found in DNA that is specific to Americans? It has been a long time since i took biology and i am guessing i missed something since then. Thanks again

    • American = Native American.

      The American component in my admixture runs is highest among the Southern Native American groups in the reference data. It is also high among Native Americans in North America.

  3. Okay, that makes perfect sense. Thanks again.

  4. Zack,

    Two questions, which I think belong in the explainer.
    a) Read thru most of your early posts and the earliest post where K=11 and you have S Asian and Onge components is the post on April 21 2011. I understand the concept of forcing the data into 11 components. What I dont get is how those components are equated to S. Asian, Onge etc.

    Does
    1)ADMIXTURE have the Reference SNP's equating to pure S. Asian Onge etc
    2)or do you create reference SNP's equating to pure S. Asian Onge and use those when you do the analysis in ADMIXTURE.

    b) I am trying to understand the relationship between the Fst Dendogram and Reich et al, phylogenic tree. I see that you say "Just remember it's not a phylogeny". Saw one comment " Would the statement "C1 S. Asian and C4 SW Asian may be similar to the AN1+ASI of Reich et al" be correct. Would that be correct. I am sure this has been asked before, please direct me to that post.

    • Don't think of the ancestral components, like South Asian, Onge etc, as pure ancestral allele frequencies. The names of the components are just mnemonics to make it easy to remember and discuss. These labels are based on which present populations that component peaks in.

      The components are computed by ADMIXTURE using a maximum likelihood algorithm. These allele frequencies do not, in general, represent any populations in the distant past. The best way to think of admixture results is in comparison with others, both the reference populations and project participants.

      • Thanks.

        Just dropped my 23andme sample at the Jersey City USPS (In NYC at the moment). So in about six weeks or so you will have another Sri Lankan participant.

      • Thanks for the explanation.
        Just dropped my 23andme sample at Jersey City USPS (I live in NYC). So in about 6 weeks should have another Sri Lankan participant.

        Sorry if you have had many of the same. Seems like my comments dont show up.

  5. Thanks for the explanation. Its starting to make sense now.
    Just dropped the 23andme sample at Jersey City USPS (I am in NYC). So hopefully in six week you will have another Sri Lankan participant.

Leave a Reply