Zika reference sequence is not always pulled from metadata genbank file. about augur HOT 5 CLOSED

Thanks for note @alliblk. I just checked this on modular augur and the issue there is resolved. In this case, when aligning, you explicitly specify a Genbank reference to align to:

from augur.

I think if the input fasta contains the reference already then --reference-name should be used instead of --reference-sequence. It doesn't look like --reference-sequence overwrites anything, but merely appends the reference sequence to the input fasta: https://github.com/nextstrain/augur/blob/master/augur/align.py#L36-L46 ?

from augur.

@tsibley ---

Thanks for digging into this. I was assuming that it overwrote given that the resulting aligned.fasta just contained a single version of PF13/251013_18 and that this version was the right length. PF13/251013_18 was in the input sequences.fasta.

Perhaps the help for --reference-sequence could make it clearer that this expects a file (Genbank or FASTA).

from augur.

I haven't poked around modular augur yet, but glad to hear that it might fix this issue. That said, I'm wondering if it might be a good idea to also hardcode removal of the non-Troesemeier et al version of PF13/251013_18 during the prepare step, since this sequence seems so off.

from augur.

This issue seems to be resolved in the latest modular augur builds. The current Zika build uses choose_genbank for the download rule and uses the explicit --reference-sequence flag for the alignment rule. This results in the correct PF13/251013_18 sequence passing through the alignment stage and into the tree building stage.

from augur.