Denovo_wf or classify_wf for taxonomic annotation of MAGs

[Christaline-George]

Hello,

I am attempting to annotate my MAGs and was wondering what would be the most appropriate workflow for this purpose? Classify_wf or Denovo_wf?

I understand the documentation specifies the following " The de novo workflow infers new bacterial and archaeal trees containing all user supplied and GTDB-Tk reference genomes. The classify workflow is recommended for obtaining taxonomic classifications, and this workflow only recommended if a de novo domain-specific trees are desired. One should take the taxonomic assignments as a guide, but not as final classifications. In particular, no effort is made to resolve the taxonomic assignment of lineages composed exclusively of user submitted genomes."

Am I right to understand that the classify_wf is the most recommended approach? I do not quite understand the de-novo approach assuming both the classify_wf and the denovo_wf would use the GTDB reference DB. I apologize if this is a trivial question.

Also, does it mean that the taxonomic assignments from the Denovo approach must be used with caution?

[pchaumeil]

Hello,

Yes, classify_wf and de_novo_wf are fundamentally different at the moment:

• The classify_wf workflow makes a best effort to determine the correct taxonomic assignment by considering the placement of a genome and RED values.

• The de_novo_wf workflow infers a de novo tree and then decorates it using functionality similar to what is implemented in PhyloRank.

If your genomes are classified down to the species level with classify_wf, you can consider this taxonomy as final. However, if classify_wf only provides genus-level (or higher-rank) taxonomy, you should follow this up with de_novo_wf. As mentioned earlier, the taxonomy produced by de_novo_wf requires additional verification (tree curation, gene tree ...).