Bug when using keys to compute lineages

[Marius1311]

When I call mc_fwd.compute_lin_probs(keys=['Ductal', 'Alpha, Beta']), I get the traceback

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-33-4d1908c3199d> in <module>
----> 1 mc_fwd.compute_lin_probs(keys=['Ductal', 'Alpha, Beta'])
      2 mc_fwd.plot_lin_probs()

~/Projects/cellrank/cellrank/tools/_markov_chain.py in compute_lin_probs(self, keys, check_irred, norm_by_frequ)
    814         else:
    815             logg.debug(f"DEBUG: Combining recurrent classes according to `{keys}`")
--> 816             approx_rcs_ = self._prep_rc_classes(keys)
    817         keys = list(approx_rcs_.cat.categories)
    818 

~/Projects/cellrank/cellrank/tools/_markov_chain.py in _prep_rc_classes(self, keys)
   1259 
   1260         lin_colors = list(np.array(lin_colors)[cat_mask])
-> 1261         approx_rcs_temp.cat.reorder_categories(remaining_cat, inplace=True)
   1262         self._lin_probs = Lineage(
   1263             np.empty((1, len(lin_colors))),

~/miniconda3/envs/cellrank/lib/python3.6/site-packages/pandas/core/accessor.py in f(self, *args, **kwargs)
     91         def _create_delegator_method(name):
     92             def f(self, *args, **kwargs):
---> 93                 return self._delegate_method(name, *args, **kwargs)
     94 
     95             f.__name__ = name

~/miniconda3/envs/cellrank/lib/python3.6/site-packages/pandas/core/arrays/categorical.py in _delegate_method(self, name, *args, **kwargs)
   2623 
   2624         method = getattr(self._parent, name)
-> 2625         res = method(*args, **kwargs)
   2626         if res is not None:
   2627             return Series(res, index=self._index, name=self._name)

~/miniconda3/envs/cellrank/lib/python3.6/site-packages/pandas/core/arrays/categorical.py in reorder_categories(self, new_categories, ordered, inplace)
   1032         if set(self.dtype.categories) != set(new_categories):
   1033             raise ValueError(
-> 1034                 "items in new_categories are not the same as in " "old categories"
   1035             )
   1036         return self.set_categories(new_categories, ordered=ordered, inplace=inplace)

ValueError: items in new_categories are not the same as in old categories

> /Users/marius/miniconda3/envs/cellrank/lib/python3.6/site-packages/pandas/core/arrays/categorical.py(1034)reorder_categories()
   1032         if set(self.dtype.categories) != set(new_categories):
   1033             raise ValueError(
-> 1034                 "items in new_categories are not the same as in " "old categories"
   1035             )
   1036         return self.set_categories(new_categories, ordered=ordered, inplace=inplace)

If you fix this, please make sure that lineages still get assigned to the right color by simply running our tutorial.

[Marius1311]

The problem has something to do with the combination of endpoints - if I just call mc_fwd.compute_lin_probs(keys=['Ductal', 'Alpha', 'Beta']) (meaning that I excluded the epsilon endpoint), everything works.

[Marius1311]

This function is really a bit tricky - if you don't know how to fix it, we can discuss that later.