Columbus was a seer

Last month I pointed to two papers on China genetics. Rereading a bit more closely, I stumbled upon a very curious PC plot. It shows the relationship of various continental populations on the first two principal components of variation genetically. Look at how Mexican Americans from Los Angeles compare to Gujarati Americans from Houston:

i-e9b9b6223f5a49ef4b8d76e44c6b31a5-columbus.png

In world wide context South Asians and mestizos can be viewed as somewhat analogous; a stable admixture between West and East Eurasian elements. Of course, the "East Eurasian" ancestry of mestizos consists of the New World descendants of Paleolithic Siberians, while those of South Asians are the ancient long resident populations of that region of the world. In a distant sense the latter are likely the root population from which the former originated, mediated by an East Asian interlude (i.e., the populations of eastern Eurasia passed through South Asia during the Out of Africa movement).

More like this

Do you mean Gujus residing in Houston or Gujus born in Houston (and does it matter)? And what is PC1 & PC2?

resident. so i assume that includes a lot of immigrants. PC1 & 2 are the two largest independent dimensions of variance which can be extracted out of the genetic genetic data. so PC1 explains 2/3 of the genetic variance, and PC2 explains 1/5. there are lot of other PC's of smaller value on down. the scaling is tricky here, look at the Y axis. the biggest variance is africans vs. everyone else, but that isn't the information they wanted to emphasize, so they rescaled it somewhat. the X axis *roughly* seems to match something which differentiates west eurasians from everyone else (a specific example would be the gene SLC24A5, where africans & east asians are one group, an west eurasians the other).

Churchill to FDR: "Would you have us treat our Indians the way you treated yours?"

By bioIgnoramus (not verified) on 06 Jan 2010 #permalink

Razib I wrote a little script that does this PCA in a sliding window along the chromosome. As it moves along the chromosome, Mexican and Gujarati populations wander around "between" Europe and Asia but occasionally one or the other jumps closer to one ancestral population or the other.

I am trying to get someone to work on identifying these "interesting" regions where Mexicans, for example, are almost completely European, etc. Fun.....

Henry

By Henry Harpending (not verified) on 07 Jan 2010 #permalink

Great example of how two dimensions lose information.

Given how different the two populations are genetically, guarantee that the third component separates them pretty cleanly.

Sorry if I'm repeating what you already knew:

To Europeans of Columbus's age "the Indies" included everything to the east of the Indus river. This included China, Japan and SE Asia. Going west from Europe Columbus expected to land on the eastern coast of China or in Japan. He did not expect to land in what we now call India. During his first voyage to America he thought that the Bahamas were Japan and that the coast of Cuba represented China. Since China and Japan were parts of "the Indies", it was natural for him to call the locals Indians.

He would have been pretty clear about the location of what we now call India, though obviously not about its culture or people. Europeans had been drawing the triangular southern Indian peninsula and Ceylon on their maps since Greek antiquity. During his voyages to America Columbus would have been very sure that he wasn't anywhere near there.