Re: 1: Your groups will only be compared against the control group by default. If you want to compare them against each other, you can use the contrast test (see this thread for a similar case). However, unless your model takes a long time to fit, it’s probably easier to just change the reference and run the model again.
Re: 2: Treatment1:After1 indeed means the treatment has produced a change in the taxa compared to the control, and it is the difference in the treated group and control group at time After1. Treatment2:Before (I think) shouldn’t exist in your results because Before should be taken as the reference Timepoint. If this is showing up, can you try setting Timepoint as a factor explicitly? Maybe the reference is getting tripped up by the interaction. After gives the difference after the intervention in the control group only.
Re: 3: As corrections go, BH is pretty lenient, so I don’t think it’s an issue with the correction. Having only a few significant hits isn’t that unusual either and it makes for a clearer story in an analysis than “a whole bunch of things changed.” However, 15,000 taxa is a ton (way more than I usually see except in enormous meta-analyses), and if most of those are very rare, they’re probably driving up your q-values because you don’t have enough power to detect a significant difference in the rare ones. How did you arrive at 15,000 taxa? I wouldn’t normally recommend filtering in MaAsLin 3, but you might consider setting min_prevalence=0.01 and min_abundance=0.0001 or something like that.
Re: 4: I’m not sure I follow - can you post the plot so I can see what you mean? The MaAsLin 3 plots are typically intended for diagnostic purposes, so it might be better to just recreate the plot in ggplot directly.
Will