Do I really need to specify which one is Covariates (control for)?

Hi all,

I am new to MaAslin2, and wish to learn more.
I have microbiome data that I which to find specific taxa that are significantly related to the factor A, B and C (Location is categorical; Factor A, B and C is continuous variable). However, because the microbiome is also environmentally driven, I would like to remove the environmental effect. So I have data similar to the attached table below. My question is: do I need to specifically remove the environmental effect (Location), if yes, how should I do this? I am using the code below (as I want to look at significant microbiome related to all the factor A, B and C):
fit_data= Maaslin2(
input_data = df_input_data,
input_metadata = df_input_metadata,
output = “result”)

Thank you.

Hi @APeh - if you include both Location and Factors A, B, and C in the model as fixed effects, you are already adjusting for the environmental effect. You can subset the final MaAslin2 results table to the
main effects of interest and re-compute the q-values to detect significant microbiome features as described here.

Hope this helps,

Hi @himel.mallick - thank you so much for your response. I have other questions and hope you don’t mind (sorry still a newbie in Maaslin)

Question 1:
since the location is categorical and has 3 levels, how should I decide which one to include in reference? (Location A, B or C)?
I assume the code is something like this:

fixed_effects = c(“Location”, “FactorA”, “FactorB”, “FactorC”),
reference = “A,B,C,FactorA,FactorB,FactorC”) #A,B,C for location?

Question 2:
If I analyse separately and still want to remove the effect of location, would the result still be the same? What I mean is instead of including all like this:
fixed_effects = c(“Location”, “FactorA”, “FactorB”, “FactorC”)
it becomes
fixed_effects = c(“Location”, “FactorA”)
fixed_effects = c(“Location”, “FactorB”)
fixed_effects = c(“Location”, “FactorC”)

Thank you so much in advance!