In layman language, a PCA shows how genetically related human populations (or individuals) are to one another. They are usually displayed in two dimensions and look like this:

However, by using some software online, you can view PCAs in three dimensions, which provides a much clearer image of human genetic diversity, unveiling many secrets that are hidden on 2D PCAs — like this:

It’s infinitely easier to identify the various human genetic clusters and clines on a 3D PCA. It’s also easier to see how human populations are a mix between five basal races: Sub-Saharan Africans, West Eurasians (‘Caucasoids’), East Eurasians (‘Mongoloids’), South Eurasians (‘Australoids’), and Amerindians.

How to view this 3D PCA

Open this online software:
https://web.archive.org/web/20201113044821/https://vahaduo.github.io/custompca/

  1. Copy coordinates listed below and paste into the “SOURCE” tab
  2. Open the “PCA PLOT” tab
  3. Hit “RUN PCA” followed by “PLOT PCA”
  4. For a better orientation, hit “FLIP” next to “X: PC1” and then “PLOT PCA” again.
  5. Hit “3D” and “2D” to switch between 3D and 2D views.
  6. If you make a mistake, simply refresh the page and start again.

Coordinates (Updated 04/22/2023):

COPY FROM HERE: pastebin.com/iVJQ7jLD (archive)

Each dot on this PCA represents the average population of a certain ethnic group. These groups have been carefully selected to fairly represent the range of variation from each major population cluster/cline.

Population clusters/clines are labeled in a way that I thought logical/useful. However, the clines could easily be subdivided. For example, South Indians (who mostly descend from Negrito-related hunter-gatherers that originally populated South and South East Asia) could be separated from North Indians and Pakistanis.

To see genetic variation in Africa (which is not visible on the default X, Y, Z axis settings) change the Y axis to “PCA5” and hit “plot PCA” again. This smushes all of the Eurasians together but shows that South African KhoiSan are a genetically distinct cluster.