**0**wrote:

I have a bacterial count matrix rows indicating bacterium species and columns indicating samples. I have also several different groupings. For example of one of the grouping: each sample can belong to one the following groups: A,B,C,D. Let's say this grouping is factor1. From all of the groupings/factors I have made a design matrix: design = model.matrix(~ factor1 + factor2 ...)

I have additionally a grouping where each sample belongs to one of E,F,G. I would like to use Limma to find out statistically significant bacteria between E,F,G without having to do pairwise-comparisons (EF,EG,FG). That is, much like anova results are interpreted. For some reason I have found only Limma contrast matrix examples for pairwise-comparisons.