ANCOM-BC q values intergroup or intragroup

[anacoluthon]

Good afternoon everyone.
I have a question regarding the q value of ANCOM-BC. I have reviewed previous posts that state that barplot-ancombc by default, shows all features, regardless of whether or not they pass significance (q<0.05) here
My question is, if that calculated significance is against the reference group, or is it between* the features of a single group (that is, intragroup significance).
If that is the case, then the output of ANCOM-BC without the q<0.05 value explicitly written during the script, simply returns those features that change logarithmically (lfc) from one group versus another?

I hope I have been clear in my doubt, which I wanted to bring up for discussion, since I did not find anyone else with that specific concern on qiime2forum.

Hugs to all

I attach a reference image, in which the second feature observed in the reference (Lachnospiraceae_ND3007_group) showed a q=0.0478, but Collinsella showed a q=1.

[lizgehret]

Hi @anacoluthon,

Welcome to the :qiime2: forum! Apologies for the delayed reply!

I'll explain what's happening in ancombc under the hood, and hopefully this will help clarify your questions.

When running ancombc, all values included in the formula are used in a design matrix (which is used for the actual linear regression). The stats provided are a result of this linear regression (i.e. LFC, p/q vals, etc), and these stats are then grouped by the terms provided in the formula to demonstrate either enrichment or depletion relative to the reference group (i.e. intercept) chosen.

Hope this helps! Feel free to reach out with any additional questions. Cheers