Reasonable number of pathways filtered away?

Hello all!

I have been working with a HUMAnN3 dataset derieved from fecal samples with ca 21,000 pathways in it. In order to filter away some of these pathways I decided to only keep pathways with a pathcoverage “score” of above 0. This resulted in only 167 pathways remaining. I fail to believe that this is a reasonable number but I can not see anything wrong with my methodology. Have I missed something obvious here? Am I wrong in using the pathcoverage file in filtering?

Any help will be appreciated!
Kind regards,

If you do a search on the forum for “pathway coverage” you will find other discussions on this topic. In short, filtering on pathway coverage in HUMAnN 2.0+ tends to be very stringent. When I am trying to reduce a set of microbiome features for analysis, I typically filter on a combination of abundance and prevalence instead. E.g. analyze features that exceed abundance A (e.g. 0.01%) in P% of samples (e.g. 10%).