Filtering out low-abundance taxa

I’ve frequently seen in literature that people often filter out low abundance taxa from the taxonomic profiling results and apparently this can improve the false positives too (e.g., see PMID: 30035235). I wonder if MetaPhlAn2 has any built-in filtering method for this or this is something that should be done manually by the user. If the latter, are there any (normalized) abundance threshold that you would recommend (for example, the paper noted above suggests to remove any taxa whose abundance is below 0.1%).

Any comments on this?