Support multiple clustering resolutions from AnnData

[pcm32]

Currently our Single Cell Expression Atlas AnnData files (such as http://ftp.ebi.ac.uk/pub/databases/microarray/data/atlas/sc_experiments/E-MTAB-6077/E-MTAB-6077.project.h5ad) will include multiple clustering resolutions and optionally a cell_type clustering field in the obs. Currently when run it I need to select the appropiate marker gene field in the uns and the matching resolution. It would be great if CellBrowser could display all the resolutions and grab associated marker genes, instead of having to pick one.

[maximilianh]

Hm, this is something that has been suggested before a few times. Matt is also for it. I'd love to add a menu option for this, like View > select cluster field. Would that work? I can't promise to work on it too soon though.

I would probably be easy to add. the only problem is to modify cbBuild to pre-compute the centroids for the cluster labels. If you don't need those, it should be very little code.

[matthewspeir]

We now have the 'label by annotation' feature which should allow us to display the labels for the different resolution.

@maximilianh I think the next step would be to allow one to specify multiple marker genes files, so that we can display the markers for each resolution?

[maximilianh]

Yes, thanks Matt for reminding me of this. Do we already have a ticket for it? (we may have one). The "markers" cellbrowser statement only would need another key in the list, "fieldName", which indicates the field for which the markers are active.

…

On Tue, May 24, 2022 at 8:13 PM Matt Speir ***@***.***> wrote: We now have the 'label by annotation' feature which should allow us to display the labels for the different resolution. @maximilianh <https://github.com/maximilianh> I think the next step would be to allow one to specify multiple marker genes files, so that we can display the markers for each resolution? — Reply to this email directly, view it on GitHub <#215 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AACL4TJDFLU7KT6JU55PZCDVLUL4XANCNFSM43BJWWWQ> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[matthewspeir]

Yep, ticket here: https://redmine.soe.ucsc.edu/issues/29299

[michael-kotliar]

Hi! Just curious, if this feature has been already implemented and we can use multiple gene markers for different resolutions?

Thanks!