Reference sequence not removed when calling augur align

[trvrb]

This should be able to be replicated by running snakemake -p results/aligned.fasta in https://github.com/nextstrain/ncov.

This calls augur align with --remove-reference. However, reference MN908947 still appears in output.

This is also replicated by running the same set of commands from https://github.com/nextstrain/zika, wherein PF13/251013_18 still appears in results/aligned.fasta.

[godotgildor]

I'm having trouble reproducing this using the Zika example.

If I run:
snakemake -p results/aligned.fasta --cores
in the zika repository, looking in the results/aligned.fasta I do not see an entry for PF13/251013-18.

If I edit the Snakefile to remove the --remove-reference and then run
snakemake -p results/aligned.fasta --cores
the resulting results/aligned.fasta includes the entry:
>KX369547.1 Zika virus strain PF13/251013-18, complete genome

Did I miss something?

[huddlej]

@godotgildor The Zika issue only occurs with the full Zika build and not with the tutorial build. See the complete description of this issue for more details.

[tolot27]

@godotgildor Please can you rerun your test with the most recent version? The issue should be fixed now.

[huddlej]

The solution to Trevor's original issue is partially data-dependent. A fix that would help from the augur side, though, is to change the name field used for reference sequences from id to name. This would be a breaking change in the sense that any build out in the world that relies on the GenBank record's id instead of name would break. However, this change would also avoid the kind of issue Trevor described and also allow users to provide FASTA references instead of GenBank files in the future.

I'm labeling this as an issue to discuss within the Nextstrain team for now. It doesn't need any further development until then.

[trvrb]

I'm going to close this immediate issue as the real issue is as @huddlej says one of naming of Genbank reference sequences. I don't think high priority.