Use "accession" column as ID column

[j23414]

Description of proposed changes

The main purpose of this commit is to ID records by "accession" to directly match changes in nextstrain/mpox@927ad6c

Additionally, the uncompressed sequence and metadata files are moved to the data folder instead of the results folder to align with the monkeypox and zika pipelines.

Related issue(s)

• This change is a prerequisite to Generalize Ingest #6

Checklist

• Checks pass
• Can run a manual check

git clone https://github.com/nextstrain/dengue.git
cd dengue
git checkout id_by_accession

nextstrain build \
  --aws-batch \
  --aws-batch-cpus 4 \
  --aws-batch-memory 7200 . --jobs 4

[jameshadfield]

Code looks good (and works!).

I don't think strain_id_field is the best name, but this follows what we do in mpx so I think consistency is the best approach here and so I'm happy for it to stay.

During testing of this it's clear that the "strain" column in the Dengue metadata needs improvement, and on the face of it I'd suggest keeping accession as the exported node name for auspice. However I'm not experienced enough in Dengue to know what to do here, so I'll leave the decision here to your judgement. Perhaps this is a data curation problem and can be fixed by future PRs? (Perhaps it's already fixed in new_ingest?) As examples:

• We have "strain" names such as 2.36E+11, 00697/11, DBS1.
• Capatilisation is irregular, e.g. DENV2 & denv2
• Sometimes / is used as the word-separator, sometimes |
• A lot of strain names are duplicates. When the dataset is loaded into Auspice you'll see error messages in the console such as Tree node detected with a duplicate name. Changing 'New_Guinea_C' to 'New_Guinea_C_670d9b' and continuing..., but the tree will still be displayed and almost all functionality will remain.

[joverlee521]

non-blocking

Noting that this is the third copy of this script (after monkeypox and rsv).

Seems like with GenBank data, this will be a common pattern. Maybe we should push on nextstrain/augur#1264 or nextstrain/auspice#1668 to solve this across the board.

[j23414]

Huh, agree that pushing the linked issue and PR would be a more streamlined solution for the future.

[victorlin]

https://github.com/nextstrain/private/issues/131

[j23414]

Thanks @jameshadfield for the review!

the "strain" column in the Dengue metadata needs improvement

Thanks for the context on how strain names affect Auspice, along with the console messages! Completely agree, yes, I'm hoping to discuss and address the "strain" column transformations in future PRs.

[j23414]

After a quick check-in with @joverlee521, I'm going to merge this for the time being. I'll plan for subsequent issues and PRs to better address outstanding discussion points.