View Full Version : Methods of measuring genetic distance/proximity between populations

05-23-2018, 12:58 PM
I am interested in the following question. Given a set of samples (or simply a genetic dataset) what is the best way(s) to quantify/measure the genetic proximity between me and samples from this dataset?

The most obvious is to run a genetic calculator like Eurogenes K36 and look at coordinates it calculates and then simply compare vectors of coordinates.
However, some components in calculators might be more related and therefore I think such way of comparing similarity/proximity is skewed.
I was also thinking about comparing PCA coordinates (e.g. Global25 coordinates by Davidski). I think this might be an interesting option but still I am not completely sure if it is better than comparing percentages produced by calculators...

Of course the most exact seem to be simply genome-to-genome comparison if the same snps are present in all genomes (or look at a subset of snps) but I am curious about other options we have.

05-23-2018, 04:08 PM
If I read you correctly that's not between populations. Between populations would be something like fixation index (https://en.wikipedia.org/wiki/Fixation_index). What you seek to do is measure distance between yourself and populations, correct? You could use a tool like this (https://anthrogenica.com/showthread.php?10813-Tool-for-K36-your-similarities-rates-on-maps).

05-23-2018, 05:06 PM
According to me using calculator results is indeed not very accurate for this. I think the best method is so far using Global 25 data with FAST as explained here :


Then use option "Multivariate > Similarity and distances indices" which will produce a Distance matrix for all populations and yourself (similarity index is by default Euclidean but you can change it).