Alright first thing first:-
The blog post you mentioned contains MDS (Multi-demensional Scaling) plots of genotype (meaning DNA) raw data of 9 population from Pakistan with 23-25 samples each -- probably collected using 23andMe.
To understand what this means you need to know what MDS plots are. MDS are essentially a non-linear dimensionality reduction technique. I am quite familiar with Dimensionality Reduction and use PCA quite often for my own work.
One easy to access source is this :-
https://ncss-wpengine.netdna-ssl.co.../Procedures/NCSS/Multidimensional_Scaling.pdf
"Multidimensional scaling (MDS) is a technique that creates a map displaying the relative positions of a number of objects, given only a table of the distances between them. "
This is a very important point. MDS takes as input the distances between the multi-dimensional data. Here that input is going to be Genetic Distance, which is an edit distance based on DNA substitution model. It then projects this information into a number of dimensions that you mention as input and you can use these plots to identify clusters.
It is noteworthy, that this information it works on is NOT the DNA itself but the genetic distance between DNA(s). This is the reason why any of the MDS plot itself can not give you answer to "how much DNA do these population share" BUT rather it tell you "how close do these data gather" also that is what you call a cluster. Obviously, you mis-understood clustering of genetic distance as absolute sharing of DNA and that gave you this weird idea that Pakistani only share 10% DNA with sub-saharans. And then likes of
@Iqbal Ali found "affirmation" to their "commonly held Pakistani believes".
This is very text book example of how NOT to do science.
BTW, what do you know about genetic distances, lets hear it. Or about clustering? I guess nothing. A hack job of copy paste can get you only so far 
