Error in utils.get_training_data when clusters_to_remove is not given

[krademaker]

Hi Alexander,

Running cell2fate on a dataset when no clusters are provided for removal gives the following error:

>>> adata = sc.read_h5ad('.../PEI6404A1.h5ad')
>>> adata=c2f.utils.get_training_data(adata,cells_per_cluster = 10**5,cluster_column = cluster_column,min_shared_counts = 10,n_var_genes= 3000, remove_clusters=None)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/env/lib/python3.9/site-packages/cell2fate/utils.py", line 297, in get_training_data
    adata = adata[[c not in remove_clusters for c in adata.obs[cluster_column]], :]
  File "/env/lib/python3.9/site-packages/cell2fate/utils.py", line 297, in <listcomp>
    adata = adata[[c not in remove_clusters for c in adata.obs[cluster_column]], :]
TypeError: argument of type 'NoneType' is not iterable

Which would be because adata = adata[[c not in remove_clusters for c in adata.obs[cluster_column]], :] in utils.get_training_data matches adat.obs[cluster_column] against a None remove_clusters.

Wouldn't it be more sensible either catch this NoneType error when no clusters have to be remove or to set the default remove_clusters in this function to empty list?

>>> clusters_to_remove = []
>>> adata=c2f.utils.get_training_data(adata,cells_per_cluster = 10**5,cluster_column = cluster_column,min_shared_counts = 10,n_var_genes= 3000, remove_clusters = clusters_to_remove)
Keeping at most 100000 cells per cluster
Filtered out 4114 genes that are detected 10 counts (shared).
Extracted 3000 highly variable genes.

works fine for instance.