search

EBISPOT / scxa_2_cxg

Apache License 2.0
1 stars 0 forks source link

Extend CxG conversion to cover - `uns` (general metadata) #9

Closed dosumis closed 3 months ago

anitacaron commented 5 months ago

It doesn't look like any data in uns is expected as in the schema.

These are the columns in uns: 'hvg', 'markers_louvain_resolution_0.1', 'markers_louvain_resolution_0.1_filtered', 'markers_louvain_resolution_0.3', 'markers_louvain_resolution_0.3_filtered', 'markers_louvain_resolution_0.5', 'markers_louvain_resolution_0.5_filtered', 'markers_louvain_resolution_0.7', 'markers_louvain_resolution_0.7_filtered', 'markers_louvain_resolution_1.0', 'markers_louvain_resolution_1.0_filtered', 'markers_louvain_resolution_2.0', 'markers_louvain_resolution_2.0_filtered', 'markers_louvain_resolution_3.0', 'markers_louvain_resolution_3.0_filtered', 'markers_louvain_resolution_4.0', 'markers_louvain_resolution_4.0_filtered', 'markers_louvain_resolution_5.0', 'markers_louvain_resolution_5.0_filtered', 'neighbors', 'pca', scrublet

anitacaron commented 5 months ago

Schema values: title, batch_condition, {column}_colors, default_embedding, X_approximate_distribution

dosumis commented 5 months ago

We may have to mine some of the other files for general metadata.

anitacaron commented 4 months ago

Only the title is required. Got this from the IDF file.

anitacaron commented 4 months ago

Add default_embedding: "X_umap_neighbors_n_neighbors_20

dosumis commented 3 months ago

Add dataset curie

e.g. SCXA:E-CURD-2

This is sufficient for rolling links to dataset files in general on FTP, SCXG web site and the CxG standard h5ad file that we produce.

TBD: Do we store this in uns or do we just make it the ID of DataSet? Decision: Save as property in dataset object.