Eurasia + Mozabites + Papuans

By Razib Khan | February 16, 2011 1:20 pm

I’m in a hurry right now, and won’t be posting much this week. But, I thought I’d dump some of the ADMIXTURE runs I have. This is one with 80,000 markers, and Eurasian populations, Papuans and Mozabites. I removed the New World and Africa to constrain the variance space. This time I’ve labelled the ancestral components, but do not take them totally literally. I think in the future I might just remove the Kalash to see what happens. This is K = 7. Not too busy, but I think enough K’s to separate out the various West Eurasian groups. Additionally I’ve put the genetic distances, Fst, below, and, visualized them on an MDS. Nothing too surprising.

Fst
Northeast Asian South Asian European West Asian Kalash Southeast Asian Papuan
Northeast Asian 0 0.1 0.13 0.142 0.137 0.052 0.225
South Asian 0.1 0 0.058 0.07 0.066 0.097 0.201
European 0.13 0.058 0 0.056 0.075 0.132 0.236
West Asian 0.142 0.07 0.056 0 0.098 0.139 0.238
Kalash 0.137 0.066 0.075 0.098 0 0.136 0.243
Southeast Asian 0.052 0.097 0.132 0.139 0.136 0 0.214
Papuan 0.225 0.201 0.236 0.238 0.243 0.214 0

CATEGORIZED UNDER: Genetics, Genomics
MORE ABOUT: Admixture, Genomics
  • http://washparkprophet.blogspot.com ohwilleke

    Given that Papuans split off ca. 45,000 years ago from Eurasians, and that Fst ought to bear some relation to period of divergence between populations, this would suggest that the Kalash, South Asian, European, West Asian cluster has diverged (or alternately diverged and readmixed) much more recently, and similarly that the Northeast Asian/Southeast Asian split is either quite recent or has readmixed concealing older divergences.

    The intensity of the divide between South Asia and Southeast Asia is somewhat surprising as well, not because it doesn’t fit with what we know from lots of other genetic studies in the past decade, but because it isn’t really obvious why was so much easier for West Eurasia to blend with South Asia than it was for East Eurasia to do so. The geographic barriers between India and China of obvious. But, it is far less obvious why Burma should pose such an imposing population genetic barrier between India and Southeast Asia.

    The fact that the Dardic speaking Kalash are closer genetically to the European than the South Asian sample, despite being located in the Hindu Kush of South Asia, is also notable — suggesting that the Kalash have a quite high genetic contribution of Indo-Aryans from the Northwest.

  • http://www.zackvision.com/weblog/ Zack

    Any particular reason for including the Mozabites?

  • Laredo

    I find it interesting that the lowest Fst distance between any of the statistical clusters/components is between the West Asian “ancestry component”, and the South Asian “ancestry component” (only 0.07). I too am often surprised at how low the levels of the “East Eurasian” components (Northeast Asian, Southeast Asian) are in South Asian populations (although the jungles of Burma are very formidable). Besides low levels of the “East Eurasian” components, all of the primary “ancestry components” found in these populations are “West Eurasian” (South Asian, European, West Asian, and Kalash). Also, with exception to Sardinians, all “West Eurasians” have some level of the South Asian “ancestry component” and some level of the Kalasha one ( Mozabites and Sardinians both being the exception in this case).

    I have always been perplexed at how the Kalash are cut up and identified in such varied ways by different researchers using these types of analysis (the identification of so-called “ancestral components” or clusters). Rosenberg et al. (2002) found the Kalash rather distinctive with 377 markers. Li et al. (2008) using 650,000 markers (only Single Nucleotide Polymorphisms) found that the Kalash were completely South Asian, having none of the European, Middle Eastern, or East Asian “ancestry component” common in other populations who also belonged primarily to the South Asian “ancestry component”. Tishkoff et al. (2009) found that using 1,327 markers, the Kalash represented a similar pattern to their “Central Asian” sample (more like northern South Asian and Central Asian). And the analysis I see here with 80,000 markers is clear to their distinctiveness at K=7.

    Similar observations can be made for the other components. This is one of the many reasons why I am inclined to view the “ancestry components” as statistical constructs that depend on the number and type of markers used, the size of the sample, the populations included, and the amount of Ks the user specifies, something which I think most would agree with. I mean, its not as if the clusters are literal ancestral populations that once roamed the earth to mix with other such populations. And anyways, we are a thoroughly mongrelized species, for where ever the trade and invasion routes have went, so have the genes. There never really has been any such thing as a “homogeneous” or “pure” population (whatever those terms may mean), and all the clusters are themselves as composite as the contemporary populations in which they are found. Back to the Kalash, perhaps a study of Afghan Nuristanis might clear things up (sadly, Afghanistan is so sparsely represented, although I assume that ethnic Afghans in Pakistan (Pathans, should be Pashtuns or Pukhtuns) are similar to their relatives in Afghanistan and do the job, along with Tajiks from Tajikistan whom I have never seen represented in general by any study). Always fascinating material.

  • http://washparkprophet.blogspot.com ohwilleke

    “This is one of the many reasons why I am inclined to view the “ancestry components” as statistical constructs that depend on the number and type of markers used, the size of the sample, the populations included, and the amount of Ks the user specifies, something which I think most would agree with. I mean, its not as if the clusters are literal ancestral populations that once roamed the earth to mix with other such populations.”

    Ancestry components in this analysis are statistical constructions, but of course, there were indeed literal ancestral populations that once roamed the earth to mix with other populations. They just weren’t necessarily the ones that the computer creates.

    In contexts like the population history of African Americans, or Mestizos in Mexico, or a large share of South Asians, an admixture of ancestral populations is quite a good model of what we know (or have strong reasons to suppose) to have happened in real life. The correspondence between ancestry components and literal ancestral populations is pretty good.

    When we have a dim idea regarding the population histories involved, we have to be more cautious. But, while autosomal genetics do a poor job of telling us anything about the sequence and relationship of different populations, uniparental genetics and analysis of small numbers of differentially selected alleles that code a functional phenotype (e.g. lactose intolerance) can give us a quite well informed sense of what sort of model we ought to try to have as a framework when we look at autosomal genetics, and tools like domesticated plant and animal genetics, continuity of archeological sites, geographical barriers in light of known paleoclimate, and language can provide hints that are worth something as hypothesises to test.

    For example, mtDNA is a good way to get a decent estimate of the number of maternal source populations (the K value) that makes sense in an admixture analysis, and is also a good way to know what data sets need to be included in the analysis to make it meaningful. If your data set is full of mtDNA C haplogroups and you put in West Eurasian whole genomes to compare to you sample, you are going to get noise and junk. If you put populations that have mtDNA C haplogroups already, you might get a sense for which ones your population bears the strongest affinity towards.

  • Laredo

    There have been literal ancestral populations, this I never did deny, and I attempted to make the same general point that you have articulated in a superior manner (“They just weren’t necessarily the ones the computer creates.”) , but we must remember that there are no panmictic populations, and there never have been any panmictic populations. The actual ancestral populations being represented as components were obviously themselves products of admixture and sub structured. In the end, any modeling system (or any model in general) necessarily involves simplification. I find that the rest of your points are shared ones, although I am less enthusiastic about uniparental markers, but on their being an important part of the puzzle, nobody can deny.

  • onur

    I find it interesting that the lowest Fst distance between any of the statistical clusters/components is between the West Asian “ancestry component”, and the South Asian “ancestry component” (only 0.07).

    It is not the lowest Fst distance here. The “West Asian” component is closer to the “European” component with an Fst score of 0.056 than to the “South Asian” component, and the lowest Fst distance here is between the “Northeast Asian” and “Southeast Asian” components with an Fst score of 0.052. BTW, the “European” component is closer to the “South Asian” component with an Fst score of 0.058 than the “West Asian” component is.

    Obviously you are ignorant of reading decimal numbers: 0.07=0.070, but the last zero was dropped because that it is redundant in writing. But you ignorantly read 0.07 (=0.070) as if it is 0.007.

NEW ON DISCOVER
OPEN
CITIZEN SCIENCE
ADVERTISEMENT

Discover's Newsletter

Sign up to get the latest science news delivered weekly right to your inbox!

Gene Expression

This blog is about evolution, genetics, genomics and their interstices. Please beware that comments are aggressively moderated. Uncivil or churlish comments will likely get you banned immediately, so make any contribution count!

About Razib Khan

I have degrees in biology and biochemistry, a passion for genetics, history, and philosophy, and shrimp is my favorite food. In relation to nationality I'm a American Northwesterner, in politics I'm a reactionary, and as for religion I have none (I'm an atheist). If you want to know more, see the links at http://www.razib.com

ADVERTISEMENT

See More

ADVERTISEMENT

RSS Razib’s Pinboard

Edifying books

Collapse bottom bar
+

Login to your Account

X
E-mail address:
Password:
Remember me
Forgot your password?
No problem. Click here to have it e-mailed to you.

Not Registered Yet?

Register now for FREE. Registration only takes a few minutes to complete. Register now »