Hi,
I'm interpreting alpha group significance results, and I came out with a question...
The comparison between all groups is made by Kruskal Wallis test, but: What test is used for pairwise comparison? I think the most recommended test for that purpose it would be pairwise wilcoxon test, but qiime reported that it was also made through Kruskal.
Thanks a lot
I also have the same question regarding pairwise Permanova test used in beta-group significance. Thanks a lot!
There's nothing wrong (to my knowledge) with running a Kruskal-Wallis test on 2 groups.
It's a PERMANOVA test, but only including 2 groups.
The pairwise element here is that each set of 2 groups is being isolated and compared against each other. This is not a paired test, where paired samples in each group (e.g., same subject sampled at 2 different time points or locations) are being compared, and requires a separate test.
Does that answer your questions?
Perfectly.
where paired samples in each group (e.g., same subject sampled at 2 different time points or locations) are being compared, and requires a separate test. -> What would be the steps to do that in Qiime2? I have a set of samples from gut and lung locations from the same patients, and I want to compare it patient by patient.
Thanks a lot @Nicholas_Bokulich
q2-longitudinal has some paired tests, confusingly called pairwise-distances and pairwise-differences.
Good luck!