Using `--reference` flag with taxon queries produces a confusing error message about BioProject accessions.

[nekrut]

Version: 18.13.0

Description:
Using --reference flag with taxon queries produces a confusing error message about BioProject accessions.

Steps to reproduce:

# Works
datasets download genome taxon 2894792 --dehydrated

# Fails with misleading error
datasets download genome taxon 2894792 --reference --dehydrated

Error message:
Error: NCBI Datasets provides access to assembled genome data by BioProject accession and no genome assemblies were found for the given BioProject.
Retrieval of SRA data by BioProject accession is not supported by NCBI Datasets.

Expected behavior:
- If no reference genomes exist for the taxon, error should say "No reference genomes found for taxon X"
- If reference genomes exist (e.g., measles has GCF_000854845.1), they should be returned

Additional context:
Same issue occurs with taxid 11234 (measles) which has a known RefSeq assembly GCF_000854845.1.

---

[ericcox1]

Hi Anton,

Thanks for opening this issue.

Using --reference flag with taxon queries produces a confusing error message about BioProject accessions.

I agree, this is a confusing error message for this situation. We are going to update this to give a more generic message that makes more sense.

Same issue occurs with taxid 11234 (measles) which has a known RefSeq assembly GCF_000854845.1.

Separately, for virus genome assemblies, we made a change to the data last year where we are no longer labeling them as "reference genomes", so the --reference flag no longer returns any genomes for any virus taxon query. In order to find RefSeq genomes, you'll need to use --assembly-source refseq as follows:

datasets summary genome taxon 11234 --assembly-source refseq --report ids_only
{"reports": [{"accession":"GCF_000854845.1","source_database":"SOURCE_DATABASE_REFSEQ"}],"total_count": 1}

We're going to add documentation that explains this.

Thanks for your feedback @nekrut

I'll leave this issue open until we make these updates.

Best,
Eric