Extend functionality of tl.rank_genes_groups()

• Additional function parameters / changed functionality / changed defaults?
• New analysis tool: A simple analysis tool you have been using and are missing in sc.tools?
• New plotting function: A kind of plot you would like to seein sc.pl?
• External tools: Do you know an existing package that should go into sc.external.*?
• Other?

1. I would be great if the reference option in sc.tl.rank_genes_groups() would accept not only 'rest' or a single group identifier, but a list of identifiers (as is possible for the groups option). Currently, lists such as reference=['g1','g2','g3'] are not accepted:

   ---------------------------------------------------------------------------
   TypeError                                 Traceback (most recent call last)
   Input In [72], in <cell line: 2>()
         1 # each WT sample individually vs all other WT samples
   ----> 2 sc.tl.rank_genes_groups(remerged_WT, groupby='sample', groups=WT_samples, reference=['g1','g2','g3'], 
   method='wilcoxon')
   
   File D:\Programme\Anaconda\envs\squidpy\lib\site-packages\scanpy\tools\_rank_genes_groups.py:570, in 
   rank_genes_groups(adata, groupby, use_raw, groups, reference, n_genes, rankby_abs, pts, key_added, copy, method, corr_method, 
   tie_correct, layer, **kwds)
   568     if isinstance(groups_order[0], int):
   569         groups_order = [str(n) for n in groups_order]
   --> 570     if reference != 'rest' and reference not in set(groups_order):
   571         groups_order += [reference]
   572 if reference != 'rest' and reference not in adata.obs[groupby].cat.categories:
   
   TypeError: unhashable type: 'list'
   

2. Also I would appreciate if sc.tl.rank_genes_groups() would allow testing against a reference which is (partly) included in the groups that are being tested. In other words, I would like to test all individual groups ['a','b','c','d'] against a fixed reference ['b','c','d']. Currently, if I test a group which is also set as reference (eg groups=['g1','g2','g3'], reference='g2'), the group (g2) is skipped and not added to adata.uns["rank_genes_groups"]. When using reference='rest', the tested group is excluded from the reference (eg reference is ['b','d'] when ['c'] ist tested).

Much appreciated!
Simon

Start with sc.tl.rank_genes_groups() in scanpy/tools/_rank_genes_groups.py and reproduce the reported list-reference TypeError. Trace how groups and reference are validated and how results are stored in adata.uns["rank_genes_groups"]. Done means list-valued references are accepted, overlapping tested groups are handled as requested, and existing behavior remains intact.