Input table used for core-features core microbiome


I want to define the core microbiome across my samples using qiime feature-table core-features. I've been trying it using the table generated from running deblur after assigning taxonomy and filtering out the rare ASVs, but I'm only getting 2-3 core features showing up. Is this due to the table I'm using?


I guess you are right.
If your samples are very heterogeneous, then it is not surprising that you are getting low number of core features.
Possible solutions (I don't know yor experimental design and goals, so it is up to you to decide how to proceed):

  • Collapse your table to taxa. Even 1 nt difference between ASVs will differentiate them. Collapsing ASVs to taxa will increase number of core features, though it will be taxa instead of ASVs.

  • Separate your table by metadata. If you have different groups of samples that are different from each other separating them can help you to find more core features, though it will be core features for different groups.

  • Define core-features to certain threshold (like use 70% (0.7) of samples instead of 100% (1.0)).

But having a low number of core-features is nor necessarily a bad thing. Again, it depends on your goals and experiment.


1 Like

You were right that I needed to collapse the table and differentiate my metadata. Thank you so much!

Hi @timanix!

This solution definitely worked to create my core-features, thank you! I wasn't sure if I should start a new topic for this, let me know if I should and I will!

I'm trying to use these core features to create taxonomic barplots from the tsv files, which isn't working because the taxa was collapsed and now doesn't match my taxonomy file (at least I think this is whats happening). Any advice on how I might proceed?

Here's the script I'm running and the error I'm getting.

qiime feature-table filter-features
--i-table skin-taxa-table-L6.qza
--m-metadata-file core-features-0.800-skin.tsv
--o-filtered-table table-core-skin-0.800.qza

qiime taxa barplot
--i-table table-core-skin-0.800.qza
--i-taxonomy taxa/classification.qza
--o-visualization taxa/taxa_barplot_core_skin.qzv

Feature IDs found in the table are missing from the taxonomy: {'d__Bacteria;p__Proteobacteria;c__Alphaproteobacteria;o__Rhizobiales;f__Rhizobiaceae;g__Allorhizobium-Neorhizobium-Pararhizobium-Rhizobium', 'd__Bacteria;p__Proteobacteria;c__Gammaproteobacteria;o__Xanthomonadales;f__Rhodanobacteraceae;g__Rhodanobacter',... etc.}

skin-taxa-table-L6.qza (216.9 KB)
core-features-0.800-skin.tsv (5.1 KB)
table-core-skin-0.800.qza (170.3 KB)
classification.qza (2.6 MB)

Yes, you are right - the issue is in the collapsed table.
Check this discussion how to get a barplot with collapsed tables.

In future, please create a new topic if the subject of the old one changed.


Thank you for this!

Unfortunately, I don't think I can make a tsv file like mentioned in the linked post as many of my core features are corresponding to multiple feature IDs (sometimes in the hundreds). Is there a way I can collapse the initial table and still retain the feature IDs? Or maybe another way to do this?

I'm open to other types of visualizations as well, I'm just trying to make any figure to represent my core microbiomes. I've been trying to make a heatmap as well, but am running into the exact same issue of not being able to use my collapsed table with my metadata. I'm getting close to my deadline and I'm really struggling with this, so any help at all would be super appreciated!

In your case instead of feature ids (ASVs) you should create a tsv file with your core taxonomy as both ids and taxon (just diplicate it), in that way it can be accepted by barplot plugin

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