[ancestral] VCF and JSON outputs differ

[jameshadfield]

Current Behavior

Given a site with a reference allele and a N (encoded in a VCF file), then that site is not included in the output JSON or the output VCF. This is expected behaviour unless we are running with --keep-ambiguous.

However if there is another [ATGC] allele at that site as well as the "N" allele, then the "N" allele is included in the output VCF. (The JSON output does not include the "N" allele.)

This causes a subtle but sizeable bug:

• The (nt_muts) JSON is used as input to augur export, and thus the "N" allele is not displayed in Auspice. This is expected behaviour.
• The VCF output is used as input to augur translate, which results in codons with "N" in them which are represented by the AA "X". This is unexpected.
• The resulting dataset in Auspice has AA mutations of "X" with no underlying nuc mutations.

Expected behavior

The VCF output shouldn't include the N allele.

[jameshadfield]

Fixed by Augur PR #1355 (commit cf531f6) + TreeTime PR https://github.com/neherlab/treetime/pull/263 (commit https://github.com/neherlab/treetime/commit/924d9ba37e9151ab27dbc5f87d7e50695f6439d4)

[jameshadfield]

Closed by #1355