Taxonomyreport doesn't work on same database

[etowahadams]

I am trying to generate a taxonomy report for a seqTaxDB using taxonomyreport.
When I try the following, however, I get it tells me that my seqTaxDB is an amino acid db, which it isn't.

mmseqs taxonomyreport seqTaxDB seqTaxDB report

Input database "seqTaxDB" has the wrong type (Aminoacid)
Allowed input:
- Alignment
- Prefilter
- Bi-directional prefilter
- Clustering
- Taxonomy

I know my seqTaxDB is valid since I can use it with taxonomyreport in other contexts successfully. How should I generate a taxonomy report from an existing seqTaxDB?

[milot-mirdita]

This wasn't a use case we thought of for this tool. We expected it would be used to process the output of the taxonomy output and since a few days ago a search result.

I can add support for this use-case, however, could you explain why this is useful? Seeing the taxonomic composition of a large database, such as the Uniprot will basically result in an enormous tree containing nearly every Taxon on existence. I could imagine this is useful for a small database, I guess.

[etowahadams]

Yes perhaps I am using taxonomyreport in an unusual way.

I wanted to cluster sequences of all the proteins with a domain from the NCBI domain database so I got a list of accession numbers of all the proteins containing that domain, made a fasta file with them (using blastdbcmd), then created a taxidmapping for those sequences. Using the fasta file and the taxidmapping, I made a seqTaxDB (createtaxdb with --tax-mapping-file option).

I then ran cluster on the database, and then taxonomyreport on the cluster database which worked great. I wanted to compare the taxonomic distribution of the clustered sequences to the original set of sequences so I tried taxonomyreport seqTaxDB seqTaxDB report. Perhaps there is a better way to do what I am trying to do?

[milot-mirdita]

I just pushed the changes to allow annotating a sequence database. However you should note that it will count every cluster member when you pass it a cluster result. So the taxonomyreport of the seqTaxDB before and after the clustering would (should at least if i didn't do anything wrong) be the same. They will differ from the original sequence set however.

[etowahadams]

I see that makes sense. Thank you for these changes!