Re-calculate point diameter for slideseq datasets

[sidneymbell]

Thanks to @pablo-gar for the bug report + suggestion.

For slide-seq seems like a free-for-all based on whatever scale the contributors decide to use in obsm[X_spatial] , which defaults to having the CELLxGENE diamater's for each point, thus creating overlapping points in some cases (see screenshot). I wonder if you guys have thought about this.
https://cellxgene.cziscience.com/e/327288b9-4e10-40ee-950a-ad18b434930e.cxg/

Each array is 3 mm in diameter
Each array has about 100,000 beads
Beads are 10 microns
And some basic math means that the diameter of each point should always be 0.003% of the diameter occupied by the full cloud of points.
Which makes sense since the total area occupied by the points will be 89% of the "cloud" circle area

[atarashansky]

If we make the point diameter smaller, would that have any negative consequences on point visibility when zooming in?

[dsadgat]

@niknak33 please close with link to point size ticket

[kaloster]

slack context

@thuang
turns out that the spot diameter is actually encoded in the data and it can vary depending on datasets -- hoorray we actually included this field!
It is here uns["spatial"][library_id]['scalefactors']['spot_diameter_fullres'] and its value is actually in pixels corresponding to the full res image, so if only the high res image is available you'll need to use the scale factor included in uns["spatial"][library_id]['scalefactors']['tissue_hires_scalef']
Would it be possible to adjust the diameter on a dataset basis?

[brianraymor]

Clarifying just in case. This issue refers to "slideseq datasets". The spot_diameter_fullres is only supported for visium. because it's a spaceranger output. Please see https://github.com/chanzuckerberg/single-cell-curation/blob/main/schema/5.1.0/schema.md#spatiallibrary_idscalefactorsspot_diameter_fullres.

[tihuan]

Oh thanks so much, @brianraymor ! That's a super important callout for sure 🙏

@pablo-gar so it sounds like for slideseq we might need to get the info elsewhere?

Thank you!