Contingency-based Filtration


I have applied contingency-based filtering to remove features that show up in 1 sample only in my features table. However, when I compare the taxa data which computed using original feature table to the contingency-filtered feature table, many phyla that present in more than 1 samples are omitted. I attached my result and coding for the reference:
qiime feature-table filter-features
--i-table table.qza
--p-min-samples 2
--o-filtered-table contingency-filtered-table.qza

qiime taxa collapse
--i-table contingency-filtered-table.qza
--i-taxonomy taxonomy.qza
--p-level 2
--o-collapsed-table cont-phyla-table.qza

The above table is taxa classification computed using original feature table; the underneath one is computed using contingency-based filtered feature table as describes in the above script.

What appears to be going on is you are applying your contingency-based filter (minimum samples) on the ASVs, before collapsing to taxa. The taxonomy collapsing will aggregate many different ASVs that share that taxonomy. So just because a phylum is found in more than 2 samples does not mean that the underlying ASVs are. The phyla that are lost after filtering appear to be found in only a handful of samples, anyway, so this seems to imply that the ASVs detected in those samples are unique to each sample, and hence filtered out. So I would say that this is working as expected.

If you want to filter out taxa that are not present in at least 2 samples, you should use collapse and then filter.

Good luck!

This topic was automatically closed 31 days after the last reply. New replies are no longer allowed.