Significance of between-group difference

Hi all,

Thanks for building the mixOmics package!
I am a new user examining the difference of microbiota between several sample groups.
I think the Koren16S tutorial will be the best reference for my case?

Despite the sample plot and clustered image plot showed clear difference between sample groups, are there any method to quantify the how the groups are separated?

I am not sure whether mixOmics can generate p-value directly in this case.
Currently, I am thinking of performing permutation significance test by the function pairwise.MVA.test() from the R package “RVAideMemoire”. Is this a valid approach?

I am also reading the threat here for more reference, but need more time to digest the content. Sorry if I make a duplicated issue.

Any advice would be highly appreciated!

hi @ivanllampy,

Yes, you can use a permutation method, such as those proposed in RVaidememoire if you want to obtain a p-value (the post you refer to would also bring additional ways of assessing your results).


Hi @kimanh.lecao ,

Thank you for your advice.
In common permutation method, like PERMANOVA, will usually associated with a corresponding dispersion test. Is there a dispersion test applicable to the final sPLDA model if I perform a permutation test? Or a dispersion test is irrelevant in this case?

Thank you!

hi @ivanllampy

I did not know about the dispersion test, but I think you could run it on the components of the PLS-DA, since you are in a Euclidean space.