In my study I compare two groups which are include in the `Group`

parameter.

I wonder if the `coef`

from Maaslin2 is adjusted for BMI if BMI is included in the fixed_effect like this: fixed_effects = c(“Group”, “BMI”).

Can I then compare the `coef`

from a model without BMI `fixed_effects = c("Group")`

and a model that includes BMI `fixed_effects = c("Group", "BMI")`

to get an idea how BMI is affecting the differences in Group?

Or should I use the `residuals`

to make such comparisons?

It’s a bit of a misnomer to say that including one covariate “adjusts” the coefficient of another covariate. Yes, the coefficient of Group will change if you add in BMI, but that’s beside the point. The point of including covariates (Group, BMI) in a model is to account for features of your dataset that explain variation in the outcome (bug abundances).

Unfortunately MaAsLin2 doesn’t allow flipping the model to use other variables as the outcome, which seems to be what you’re interested in doing. That’s among the design goals we’re considering for future versions, but currently it’s not possible with MaAsLin.