I find some difficulties in fully understanding how maaslin2 associates microbial composition to metadata. I understand it use mixed models with fixed and random effects, but not how it works specifically.
Lets assume the I have 4 metadata features, and I set 2 as fixed and 2 as random effects:
1)Does it choose any time different metadata as the outcome variable? use the other 1 as fixed and the other 2 as random? If so, does it choose only from the fixed effects?
2)How does it calculates the p value?
3) If one of the metadata is not set as neither fixed nor random, does maaslin want use it to model?
I actually hadn’t heard of Maaslin2 until this post and looking up their resource page sounds like there is still no manuscript or pre-print on the tool. It will probably be faster to get answers to your questions on the MaAslin2 forum.
Keep us posted if you do get your answers, might be helpful to other users!
Oh, and welcome to the forum!