[TheiaViral] Adding segment information to consensus fasta headers for flu sequences

[shelby-bennett]

🆒

📌 Explain the Request

We had a user request for the addition of the segment information to be added to the ivar.consensus fastas generated when running TheiaViral on flu samples. The desired output would be similar to how IRMA adds the information in TheiaCov. Currently, the fasta headers have the refseq accession id appended to the end of the sample name in the header which then requires users to search for the accession in NCBI to determine the segment type.

Two suggestions were made which could be implemented:
Current implementation:
>sample_name_NC_026431.1

1. Append the segment abbreviation to the end of the header:
   Suggested implementation:
   >sample_name_MP
2. Append the reference sequence header in place of the reference sequence name:
   >sample_name_NC_026431.1 Influenza A virus (A/California/07/2009(H1N1)) segment 7 matrix protein 2 (M2) and matrix protein 1 (M1) genes, complete cds
   NOTE: with this addition, the consensus files will more than likely begin to clash with the max character length allowed for vadr (52 characters). This approach will require the addition of the --noseqmaxname to the organism parameters wdl.

I've attached two example fastas for reference, but feel free to let me know if you need any additional information.

theiacov.irma.consensus.fasta.txt
theiaviral.ivar.consensus.fasta.txt

[xonq]

This may require some sort of alignment-based approach as I think it is difficult/impossible to acquire the segment names from RefSeq.
IF the segment metadata can be acquired, then it would have to be added to the header prior to Skani database construction instead of during TheiaViral. This way TheiaViral can be static and validated.

[xonq]

Looks like we can use the metadata. We have to individually extract metadata for each segment, so for update_theiaviral_db.py we will have to add a module to parse RefSeq-acquired accessions for the metadata in the header/the gene ID on each scaffold. If there are multiple genes, we will have to build a mapping for common segment lineages. Currently, Hantavirus and Influenza are the only ones our users have reported analyzing, but we want a generalized solution that is broadly compatible.