fetch-from-ncbi-virus does not include nucleotide sequences

[joverlee521]

Current Behavior

Previously, the nucleotide sequence per record would be included as sequence, since we are pulling the nucleotide sequence as part of the NCBI Virus URL

https://github.com/nextstrain/ingest/blob/c97df238518171c2b1574bec0349a55855d1e7a7/ncbi-virus-url#L78

However the monkeypox ingest workflow has been returning an empty values for sequences.
Looking back at previous versions of s3://nextstrain-data/files/workflows/monkeypox/genbank.ndjson.xz:

2023-09-05 (version id c.cdLtg8OxV1Pyl8SSlWE1_dqKpQBT.z) - still included sequences for all records
2023-09-06 (version id PaqGNfdlQXH7eV9b.WVpaOm5ioQ1pVD2) - 240/6751 records did not include sequence
2023-09-07 (version id nImnSdA8NDGCJdVuDuMmsoFB_hveCkCC) - 6071/6762 records did not include sequence
2023-09-08 (version id UZ9VwlVMqVfAeP0sMux9qE4H1e6dGZRP) - none of the 6809 records included sequences
2023-09-09 (version id VWxHnqlAUVEGRU4_ngYsJuctK7Tftyyn) - none of the 6807 records included sequences

I had wondered if there was a bug in the centralized ingest script, but running the recently deleted monkeypox fetch-from-genbank script returns the same results without sequences.

NCBI Virus observations

The nucleotide sequence field name has not changed since you still download the sequences in a FASTA file with the same field name:

https://www.ncbi.nlm.nih.gov/genomes/VirusVariation/vvsearch2/?fq={!tag=SeqType_s}SeqType_s:("Nucleotide")&fq=VirusLineageId_ss:(10244)&cmd=download&sort=SourceDB_s desc,CreateDate_dt desc,id asc&dlfmt=fasta&fl=AccVer_s,Definition_s,Nucleotide_seq

However downloading as CSV or JSON format results in empty column for Nucleotide_seq.

[joverlee521]

If the nucleotide sequences are only available for the FASTA file download, we could change the URL to download the FASTA format and then parse the FASTA file into NDJSON. However, I have not tested if all of the fields we currently download are available for the FASTA file download.

[joverlee521]

The NCBI Virus API has worked ~well for us the past 3 years, but this change really makes it less appealing to work with.
With the switch over to NCBI Datasets in monkeypox and rsv, the only missing fields from the NCBI Virus API are title and publications which we do not generally use in our pipelines anyways.

I think we can drop this route for NCBI data and just recommend the NCBI Datasets CLI.