Filtering Frequency of Features post-taxonomy?

Hi there,
Is it possible to filter the frequency of features as percentages? I have samples that have feature frequencies under 1% and wondering if I can filter these post-taxonomy?

I used this to filter OTU's beforehand:

qiime feature-table filter-features
--i-table table-trim-97-clones.qza
--p-min-frequency 5
--p-min-samples 2
--o-filtered-table filtered-table-clones.qza

And then chimera filtering using the Uchime plugin. After that, I did taxonomy calling using the 97% identity UNITE eukaryote database.

I then use this to exclude certain taxa from the barplot:

qiime taxa filter-table
--i-table filtered_seed_taxonomy.qza
--i-taxonomy vsearch-taxeuk-seeds.qza
--p-exclude "Unassigned,k__Viridiplantae,k__unidentified"
--o-filtered-table filtered_seed_taxonomy.qza

As you can tell, there are a lot of taxonomy calls that are below 10%, some are even below 1% and those cannot be seen on the bar graph at all.

Is there any other way to filter after taxonomy is done or should I bump up --p-min-frequency to a higher number and re-run taxonomy calling? Just curious if there's a way to filter frequencies/features post-taxonomy that doesn't involve me writing out an exhaustive list for --p-exclude
Thanks so much in advanced!

Hello @lucky_endophyte,

Whether or not you've taxonomically classified your sequences, your feature table has the same contents. Thus you can filter your feature table as normal. Take a look at qiime feature-table filter-features-conditionally to filter using proportions instead of counts.

1 Like