load_collections not loading variables in participants.tsv?

[jdkent]

cannot access the 'group' variable in 'participants.tsv' in the narps dataset https://github.com/jdkent/model-zoo/tree/typical/narps

this line

this output:

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
Input In [60], in <cell line: 2>()
      1 get_ipython().run_line_magic('pdb', '')
----> 2 ds1_specs = ds1_node.run(sub_contrasts, group_by=ds1_node.group_by, **next_node.children[1].filter)

File /opt/conda/lib/python3.9/site-packages/bids/modeling/statsmodels.py:444, in BIDSStatsModelsNode.run(self, inputs, group_by, force_dense, sampling_rate, invalid_contrasts, **filters)
    442 # group all collections and inputs
    443 all_objects = inputs + collections
--> 444 groups = self._build_groups(all_objects, group_by)
    446 results = []
    448 for grp_ents, grp_objs in list(groups.items()):
    449 
    450     # split group's objects into inputs and collections

File /opt/conda/lib/python3.9/site-packages/bids/modeling/statsmodels.py:347, in BIDSStatsModelsNode._build_groups(objects, group_by)
    345 missing_vars = list(group_by - set(df.columns))
    346 if missing_vars:
--> 347     raise ValueError("group_by contains variable(s) {} that could not "
    348                      "be found in the entity index.".format(missing_vars) )
    350 # Restrict DF to only grouping columns
    351 df = df.loc[:, list(group_by)]

ValueError: group_by contains variable(s) ['group'] that could not be found in the entity index.

the simple addition of entities and collections is not the behavior we want:

pybids/bids/modeling/statsmodels.py

Line 443 in 993362b

all_objects = inputs + collections

collections contain the variables (e.g., group), but the filename does not contain the information required.