Compatibility of SEPP Silva128 with Taxonomy Classification Using Silva138

User Support
1

Hi QIIME 2 Community,

I’m currently working on a microbiome analysis project and have encountered a situation where I'm using different versions of the SILVA database for different steps in my workflow. Specifically, I am using SEPP with the SILVA 128 reference database to build my phylogenetic tree, but I am assigning taxonomy with the SILVA 138 classifier.

Here’s a summary of my workflow:

  1. Phylogenetic Tree Construction:
  • I used SEPP with the SILVA 128 reference database (sepp-refs-silva-128.qza) to insert my representative sequences into the phylogenetic tree.
  • After running SEPP, I filtered my feature table based on the sequences that were successfully placed in the tree.
  1. Taxonomy Assignment:
  • For taxonomy classification, I used the SILVA 138 database (silva138_rhizosphere.qza) to assign taxonomy to my sequences.

My Question:

Is there any potential issue with using SEPP Silva128 for the phylogenetic tree but Silva138 for taxonomy assignment? Specifically:

  • Could the mismatch between the SILVA versions cause any significant loss of data or misclassification during the SEPP filtering step?
  • Are there potential taxonomic inconsistencies I should be concerned about when comparing the results from these two different database versions?

I’m particularly interested in understanding how this might impact analyses focused on different taxonomic levels (e.g., genus vs. species).

Thank you in advance for your insights!

2 Likes
2

I am also interested in this question!

1 Like
3

Hi @Siyu_Song ,

This is a really great question. The short answer is no, there should not be an issue, but as always it depends.

It depends on whether you will need to directly compare results generated by these two resources. In a standard 16S workflow the phylogeny is usually only used for diversity analyses, and the ASVs (not the taxonomy) are used as tip labels to relate this to your feature table. So the different database versions should not cause an issue.

This should not cause any incompatibility issues, but indeed the SILVA 128 release is quite a few years old and so might not be as complete as the latest version, so there is a small chance that you might fail to insert (and hence filter out) some ASVs into the tree. But this seems unlikely, as the point of using SEPP (vs. a tree mapping approach) is to insert new sequences into a tree.

No I think not, because you are just using the SILVA 128 tree as a backbone to splice in your own ASVs, and the tips are then labeled by those ASV IDs. I suppose if you want to annotate/color the tree by taxonomy this is where inconsistencies could cause issues (as the output tree would contain both your ASVs and the reference sequences), but I am not sure.

I hope that helps!

3 Likes