Perspectives using traditional unsupervised admixture analysis

Trying to border us closer on our reality and on the side of believable I have done the following series of Admixture analyses using samples of living Eurasian populations.  My database consists of 0.5 million SNP's over the world, but to minimize the effect of recent genetic drift I made a LD pruning operation decreasing the amount of SNP's to 125 kSNP's. Without that many isolated sample groups would hijack the analysis and turn the history backward.   But there are many kinds of genetically isolated populations;  the Basques are not a young isolation.  Conversely in large areas in the North Europe and in the North Asia many populations are very young and they can make the history really backward in genetic tests. Those small northern, not so long ago diverged groups, were hunter-gatherers or nomads without borders.  Once western or eastern civilizations with text books appeared to the neighborhood they were localized and they we named.  This happened only a few hundreds years ago.  No one really know where they lived and where they came from THOUSANDS years ago.    

This post just to remind about the present-day reality after hypocrite uniparental speculations about our ancestry seen on the internet and in popular science.  The command line was

$admixture1.3 data.bed n -j4 -B

where n is 9-15 and 20 (20 only in numeric data) .

You can repeat this test with your own data and post results here.  Remember to carry out LD-pruning to get rid of the misleading recent genetic drift.

The data includes 560 individual,  20 samples at its peak in each group.  I computed group averages to make the results more easy to read.  The data behind bars can be downloaded here.

Be aware of the bar colors, because I had not nerve to arrange colors manually and automated "R barplot" coloring was not a bullet proof solution and same colors exist repeatedly with several k values. You can check the result using downloaded data.