Identifying absent OTUs

I see significant differences in the number of OTUs present between two treatments. How can I generate a list of which OTUs are absent? I’m working with the OTU table output after taxonomic assignment with the Greengenes database (per the Moving Pictures tutorial).

Thank you!

Hi @slatterm,
Are these differences the output of a test, such as ANCOM? If so you can simply use the output from ANCOM that has the name of the OTUs in it and use that list to filter out those features from your rep-seqs or feature-table as you need.
See the filtering tutorial here for a few different options on how you want to go abouts this.

2 Likes

Hi, thanks for the reply. The coremetrics showed that Treatment B had significantly fewer Observed OTUs than Treatment A.

I’m trying to get a list of which OTUs are absent from Treatment B. If I understand correctly, ANCOM is meant for calculating differences in abundance, but I’m not sure that’s the best tool to use if all I want is a list of OTUs that are missing from Treatment B relative to treatment A.

1 Like

Hi @slatterm,
Oops, sorry my original answer was a bit off your original request. I should read these a bit more carefully. Ultimately, I think we’ll still need to use the filter plugin but let’s try this approach.

First use the metadata filtering method to create 2 new feature tables, one that only has Treatment A samples, and one that only has Treatment B.
Then simply filter your original representative sequences (rep-seq.qza) using these tables with the filter-seqs plugin. So:
rep-seqs.qza - Treatment A = a list of features that are only in Treatment B
rep-seqs.qza - Treatment B = a list of features that are only in Treatment A

I believe this should get you what you are looking for.

2 Likes

I was able to create new feature tables for Treatment A and Treatment B, but I’m not clear on how to subtract those tables from my rep-seqs table using filter-seqs.

If the filtered feature table for Treatment A is called A-table.qza, is this what you’re suggesting to get a list of features that are only in Treatment B:

qiime feature-table filter-seqs
–i-data rep-seqs.qza
–i-table A-table.qza
–o-filtered-data B-rep-seqs.qza

?

Thank you for your help thus far!

Hi @slatterm,

Perfect!

Exactly.
From the filter-seqs help file:

 --i-table ARTIFACT PATH FeatureTable[Frequency]
                                  Table containing feature ids used for id-
                                  based filtering.  [optional]

So your newly filtered tables should be accepted here and will basically just use the IDs in it to filter from your rep-seqs.
Let us know how it goes!

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