Can I apply log transformation on proportions in compositional microbiomic data

Question

[mario_de_manner]

Hello,

I have an abundance table with several samples as output from MetaPhlan, a classification tool for shotgun metagenomics data. From how the program is written, it can only tell you relative abundances or proportions of taxa and not absolute counts. However, I would like to test the R package selbal, which searches for microbial signatures in the dataset to explain phenotypes of interest. From how I understand the software applies log transformation on the data.

Since my data is zero-inflated and proportional, can I run selbal on the data? Or will I introduce statistical bias by doing so?

Thanks a lot,

Philipp