All the little things...

that keep me from writing.

Related issues: #341 (closed), #342 (closed), #339 (closed)

TODO:

• notebook 1: add option to choose batch column name
• notebook 3: add gene_column param to predict_cell_cycle
• notebook 3: store normalization and batch layer -> !395 (merged)
• group marker: tools.marker_genes.run_deseq2 -> ValueError: Observations annot. 'obs' must have as many rows as 'X' has rows (3), but has 42 rows. moved into issue (#371 (closed))
• group marker: pl.embedding.feature_per_group -> memory leak (has to do with the matplotlib renderer I presume) seems related to hexbin style moved into issue (#372 (closed))
• group marker: run_rank_genes don't use top_n it should rank all genes not the top n! top_n is None (=all) on default
• error when SCSA can't assign a cell type to a cluster
• upper/lower case inconsistency with marker repo
• expose marker_repo min_hits parameter to allow small marker lists
• gsea notebook: unused prerank parameters (threads, min_size, max_size, permutation_num); already solved on dev
• annotation: option to choose/delete/rename annotation columns
• SCSA is run multiple times with the same input. two_column, score, UI are all the same as it ignores all scores
• the column naming scheme of the annotation notebook sometimes includes the date e.g. MR_symbol_20250324104054_leiden_1.1 make that optional. It happens because some files are stored. Running a second time will add the date.
• gsea: auto-calculate gene ranking if needed
• ATAC notebook 1: Add "None"-option # Either a single column (str), a list of three columns (['chr', 'start', 'end']) or None to use index.
• ATAC notebook 1: change param to prepare_atac_anndata(h5ad_path=path_h5ad if isinstance(path_h5ad, str) else None)