feat(website): show Jaccard index for predefined Nextclade variant mutations

[fhennig]

resolves cbg-ethz/WisePulse#240

Compute and display a Jaccard index column for mutations loaded from Nextclade tree collections, using the same clinical LAPIS queries as the computed signature mode. When no clinical sequences match the lineage (e.g. no public data available), the column is silently omitted.

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
dashboards	Ready	Preview, Comment	Jun 25, 2026 1:51pm

[fhennig]

Note to self, I haven't checked this yet.

[fhennig]

I see a few issues:

• The Jaccard index isn't rounded to 2 digits, it looks weird - why isn't it rounded, is there code duplication?
• From a UX perspective, I find it confusing, where is this number coming from?
• If I Pick a variant like A.11, the column is just missing. There should be an info box

I'm thinking, maybe we could make an info box specifically titled 'Jaccard Index Information' and it would contain partially the information that is in the current box (how many clicincal seqs are there for the selected lineage?) I think that would help with the disconnect I'm now feeling

[fhennig]

We have this now ... Hmm ... There are 100+ mutations, and we can't sort by relevancy, which is annoying.

i will try and see if we can implement the threshold at least.

... Hmmm with thresholding I don't see any relevant mutations anymore. very weird. I think I'll have to look at this differently.

---

I found these that seem with a "high" score:

In the other mode the scores are much higher. Something seems quite wrong with the computation.

I think I first need to reason about it myself: How is this computed? Since we're using two different data sets, it's easy to mix things up.