Stochastic build failure due to lack of gene coverage

[joverlee521]

Error from iqtree during augur tree:

[batch] [2024-12-31T19:12:26+00:00]   ERROR: Some sequences (see above) are problematic, please check your alignment again

Log file results/a/F/all-time/F_aligned-delim.iqtree.log contains warning:

WARNING: Sequence KX510248 contains only gaps or missing data

[joverlee521]

Checked KX510248 in the input alignment file results/a/F/all-time/F_aligned.fasta and it's only Ns.

[joverlee521]

Looking at the GenBank record for KX510248, it has a gap 4946..7456, so it's missing the entire F gene (5697-7421). I would think the sequence would get filtered out by the 0.3 min_coverage filter. However, looking at the metadata.tsv, the F_coverage for KX510248 is 1.0, so we must not be taking Ns into account when calculating coverage.

[joverlee521]

Ah, the gene coverage columns are added during ingest by extend_metadata.py, which compares for the alignment start and end to the gene coordinates, but does not account for gaps/Ns. I remember now that this is why dengue took the different route of calculating gene coverage from the Nextclade's translation output.

I'm going to leave this as-is for now as it's only a stochastic failure that only happens when certain sequences are included in the build, but in the long term, we should revisit having Nextclade calculate the cds coverage.