Pairwise-distances and pairwise-differences


(Solveig Tangedal) #1

Dear all,

I was so happy to see the changes made to pairwise-differences a while back, where --p-group-column was made optional. Now I wonder if the same change could be possible for pairwise-distances? It allows for pre/post treatment analysis when the --p-state-column separates two groups to be compared.

Best wishes,
Solveig


(Evan Bolyen) #2

(Nicholas Bokulich) #3

Thanks for the suggestion!

I was mulling this when I made group-column optional for pairwise-differences. I decided against it because then there is nothing being compared in the visualization, and the statistical tests are null. Making this column optional was much more urgent for pairwise-differences since that still performs a statistical test even on a single group!

As a workaround, it is also possible to download the paired distances as a TSV in that visualization, so you can just input some other metadata column as the group-column and then download and plot the total distribution.

I have raised an issue for this feature but it will probably take a while to get around to this — contributions are always welcome if you would like to contribute these changes to the code :smile:


(Nicholas Bokulich) #4