Question about fix_formula

[afedynak]

Can someone clarify the parameter 'fix_formula' and how it should be different than 'group'

[FrederickHuangLin]

Hi @afedynak,

fix_formula is the character string expresses how the microbial absolute abundances for each taxon depend on the fixed effects in metadata. For instance, if you are interested in the impacts of age and bmi, you can specify fix_formula = age + bmi. If your variable of interest is continuous, such as age, that's it, you do not have to specify the group argument.

However, if your variable of interest is categorical, for instance, bmi, which contains lean, overweight, and obese categories. You may want to specify both fix_formula = age + bmi and group = bmi to let ancombc or ancombc2 function know that you are interested in performing group comparisons (pairwise test, trend test, etc). group argument is optional unless the multi-group comparison or presence/absence test is desired.

Best,
Huang

[ElizabethOLeary]

Hello Huang,

I have a similar question to the above, except I have two categorical variables of interest, "Treatment," which can be either "Placebo" or Probiotic," and "Diet," which can be either "Standard" or "Western."

Presently, I am unable to find a way to run the ancombc2 function such that it will perform the multi-group comparison. Is there a way to do this with two categorical variables?

Thank you for any insight you can provide!

Just FYI, here is my code (it will only run with one categorical variable in "group". If I run group = 'Treatment+Diet' I get the following error message: Error: subscript contains invalid names.

out_ta6_txd = ancombc2(data = physeq, assay_name = "counts", fix_formula = 'Treatment+Diet',
tax_level = "ta6",
p_adj_method = "holm", prv_cut = 0.10, lib_cut = 1000,
group = 'Diet', struc_zero = TRUE, neg_lb = TRUE,
alpha = 0.05, global = TRUE,
n_cl = 1, verbose = TRUE)