theislab / ncem

Learning cell communication from spatial graphs of cells
https://ncem.readthedocs.io/
BSD 3-Clause "New" or "Revised" License
103 stars 13 forks source link

extensions to support supervised project #69

Open SabrinaRichter opened 2 years ago

SabrinaRichter commented 2 years ago

PR Checklist

Description of changes

Technical details

Additional context

github-actions[bot] commented 2 years ago

Hi @SabrinaRichter,

It looks like this pull-request is has been made against the theislab/ncem master or main branch. The master/main branch should always contain code from the latest release. Because of this, PRs to master/main are only allowed if they come from any theislab/ncem release or patch branch.

You do not need to close this PR, you can change the target branch to development by clicking the "Edit" button at the top of this page.

Thanks again for your contribution!

SabrinaRichter commented 2 years ago

@AnnaChristina the supervised manuscript works on binary cell types for schuerch (https://github.com/theislab/tissue/blob/b53bcd6bbeefb10f2930fc5c55e51d6d4cfefa4c/tissue/data.py#L1641) any idea how you want different coarseness levels being handled? multiple cluster_col_xxx columns and then optionally selecting another cluster_col for node_type one-hot encoding set to 'cluster_col_preprocessed' by default?

AnnaChristina commented 2 years ago

I would define this as an additional column in obs of celldata, and I'll check how we ensure this is then subsequently used.

Yes would also go for the set default als many plotting function in ncem require the preprocessed cell names

SabrinaRichter commented 2 years ago

see my edit above?

AnnaChristina commented 2 years ago

Yes also just edited!

SabrinaRichter commented 2 years ago

Can you check out the latest commit? should not have any effect on your code. currently only one level is saved in obs, but one can select which one in the init method

AnnaChristina commented 2 years ago

Looks good to me.