glm on Aldex2 - formula vs metadata columns

Hi @dgiguer and all,

Thanks for making these tools available to this community!

I am trying to use the -p-test ‘glm’ option on q2-aldex2 but I noticed there’s no option to pass a formula, rather than a single metadata column. I did notice this discussion on here, so I’m generally wondering if there’s a new version coming up, or, alternatively, an available workaround. I suppose running it through R would be one…

Thanks,
Gil

1 Like

Hi @Gil_Sharon

We are still working on implementing the glm feature into the qiime package. Currently the work around is to pass it through the R package. Thanks!!

Cheers!

Dan

2 Likes

Hi @dgiguer,

Another quick question, re p-values Aldex generates - I ran the analysis on ‘glm’ using the R package and all looks pretty good. BH corrected p-values look very (!) low at times, and at times are significant even when the plotted data doesnt look very convincing when plotting relative abundance. One example had 1-2 orders of magnitude lower than one would get running Mann-Whitney or Kruskal-Walis on that feature alone (and without taking random effects into account).

I assume I’m missing something in how you do the analysis, but wanted to make sure I’m running this in an appropriate way. I’m happy to share files and code here or offline, if you’d like to reproduce.

Thanks!
Gil

1 Like