MaAsLin2 and crossover study design?


I noticed there was one question asking something similar to this from a few years ago, but I wanted to see if anyone had any new information regarding MaAsLin2 on a 2x2 cross over design. Is there a way to account for sequence (AB/BA) interactions along with the fixed effects? I’m having issues finding information on this regrading microbiome studies.


It would be hard or impossible, depending on the model you’re trying to fit. MaAsLin2 was not designed with crossover studies in mind, so you’d need to cast your design into the limited mixed effects framework that MaAsLin2 can handle. MaAsLin2 only implements random intercepts e.g. (1|subject) and can’t do random slopes (treatment|subject) or multiple grouping factors (treatment|subject:sequence), which I think you would want for a crossover design.