Use lineages for build targets

[trvrb]

We have a number of pathogens where the primary split is in terms of geography. For example: ncov/gisaid/global/ vs ncov/gisaid/north-america/. However, for WNV it looks like there are really very distinct lineages (that often have geographic correspondence). Here's a quick phylogeny from https://www.mdpi.com/1999-4915/15/1/35:

I think you have a dengue like situation where a single phylogenetic analysis across all WNV lineages (or dengue serotypes) may be challenging to have very accurate molecular clock.

Rather than cuts between a global analysis and a north-america analysis, you might be better served by cutting by lineage instead. This would give URLs of wnv/all-lineages, wnv/lineage-1, wnv/lineage-2, etc... I wouldn't worry about trying to preserve that historical WNV/NA endpoint.

NCBI doesn't seem to have lineage well annotated, however if you source data from https://pathoplexus.org/west-nile/ you should get lineage for every sequence.

[trvrb]

Nice to see the lineage calls at https://nextstrain.org/staging/WNV/genome?c=lineage, I'd suggest to make the structure of the phylogenetic outputs follow:

• nextstrain.org/wnv/all-lineages (basically what's currently at WNV/genome)
• nextstrain.org/wnv/lineage-1A
• nextstrain.org/wnv/lineage-2

If a lineage is too rare / too small it's okay for it not to be listed out as a specific build endpoint (so starting with just 1A and 2). These should still be viewable on the full phylogeny.

Get this working well for the genome view and later on we could update this to

• nextstrain.org/wnv/all-lineages/genome
• nextstrain.org/wnv/all-lineages/env (or whatever)

But just get things all the way working with genome as implicit before worrying about splitting it out.