feat: Lists of mutations in metadata file

[theosanderson]

While thinking about casing-discussions I realised that we don't support something that is quite popular in metadata files: lists of mutations (nucleotide and amino acid) from reference as a "metadata" column. This is pretty useful in the early days of an outbreak.

In terms of implementation, I'm not sure if this is supported in LAPIS and if not, if there is appetite for it. If not it could be another argument for some sort of proxy one day. Obviously all of this is very post MVP.

[chaoran-chen]

LAPIS does not support the inclusion of mutations on a per-sample basis but if we want this, wouldn't it be very easy to add the two corresponding fields to the metadata and let the preprocessing pipeline generate the values?

[theosanderson]

Yes good point - it would be a kind of weird situation because we'd be getting the metadata to display on the sequence details page from one source (LAPIS's mutation store) and the metadata in the bulk download from another (LAPIS's metadata store) and I guess would increase memory usage but definitely a possible option

[theosanderson]

(Either the preprocessing pipeline or the export for SILO)

[chaoran-chen]

Good point. Maybe we should indeed consider adding a corresponding feature to LAPIS. Would you like to create an issue as it's your idea? :)

[chaoran-chen]

How would you handle multiple segments or genes? Should it only have one column with mutations on all segments or would it be useful to be able to only get data for selected segments? If the latter, what would a good API look like?

[theosanderson]

Thanks for considering. I made GenSpectrum/LAPIS#896 for starters. Sorry, not much thoughts on any API yet - anyone feel free to chime in!