Allow wildcard names for model variables

I’m trying to pass variable names to a bids model with wildcards like so:

“Model”: { “X”: [ ‘stimulus_type.neutral_anticipation’, ‘stimulus_type.neutral_feedback’, ‘stimulus_type.loss_anticipation’, ‘stimulus_type.avoided_loss_feedback’, ‘stimulus_type.gain_anticipation’, ‘stimulus_type.gain_feedback’, ‘stimulus_type.missed_gain_feedback’, ‘stimulus_type.loss_feedback’, ‘trans_', 'rot_’, ‘cosine*’, ‘a_comp_cor_', 't_comp_cor_’, ‘aroma_motion_', 'censor_’ ], },

This is throwing a key error trans_*

Traceback (most recent call last):
  File "/data/nielsond/python/envs/fitlins_dev/lib/python3.6/site-packages/nipype/pipeline/plugins/multiproc.py", line 69, in run_node
    result['result'] = node.run(updatehash=updatehash)
  File "/data/nielsond/python/envs/fitlins_dev/lib/python3.6/site-packages/nipype/pipeline/engine/nodes.py", line 473, in run
    result = self._run_interface(execute=True)
  File "/data/nielsond/python/envs/fitlins_dev/lib/python3.6/site-packages/nipype/pipeline/engine/nodes.py", line 564, in _run_interface
    return self._run_command(execute)
  File "/data/nielsond/python/envs/fitlins_dev/lib/python3.6/site-packages/nipype/pipeline/engine/nodes.py", line 649, in _run_command
    result = self._interface.run(cwd=outdir)
  File "/data/nielsond/python/envs/fitlins_dev/lib/python3.6/site-packages/nipype/interfaces/base/core.py", line 376, in run
    runtime = self._run_interface(runtime)
  File "/gpfs/gsfs11/users/MBDU/midla/notebooks/code/fitlins/fitlins/interfaces/bids.py", line 235, in _run_interface
    analysis.setup(drop_na=False, desc='preproc', **selectors)
  File "/gpfs/gsfs11/users/MBDU/midla/notebooks/code/pybids/bids/analysis/analysis.py", line 89, in setup
    b.setup(input_nodes, drop_na=drop_na, **selectors)
  File "/gpfs/gsfs11/users/MBDU/midla/notebooks/code/pybids/bids/analysis/analysis.py", line 213, in setup
    transform.Select(coll, X)
  File "/gpfs/gsfs11/users/MBDU/midla/notebooks/code/pybids/bids/analysis/transformations/base.py", line 87, in __new__
    return t.transform()
  File "/gpfs/gsfs11/users/MBDU/midla/notebooks/code/pybids/bids/analysis/transformations/base.py", line 223, in transform
    self._clone_variables()
  File "/gpfs/gsfs11/users/MBDU/midla/notebooks/code/pybids/bids/analysis/transformations/base.py", line 123, in _clone_variables
    for v in self.variables}
  File "/gpfs/gsfs11/users/MBDU/midla/notebooks/code/pybids/bids/analysis/transformations/base.py", line 123, in <dictcomp>
    for v in self.variables}
  File "/gpfs/gsfs11/users/MBDU/midla/notebooks/code/pybids/bids/variables/kollekshuns.py", line 184, in __getitem__
    return self.variables[var]
KeyError: 'trans_*'

I get the same thing if i use trans_.* I’m pretty sure this is happening because no matching is being done on X here: https://github.com/bids-standard/pybids/blob/15aa08fb2b146e3fcc62f359ea4a4d810d38591a/bids/analysis/analysis.py#L212-L213

[tyarkoni]

What’s the objection to just having standard globbing behavior again? Is it just to guard against accidental overinclusive queries? I guess I don’t really see that as being our concern… if someone doesn’t realize that doing a* is going to end badly, they’ll learn soon enough from experience. I would rather have the simplicity of saying that * and ? behave as they do in the unix shell.

[effigies]

We should put it in the spec. Otherwise, there’s really no good way to specify a potentially variable number of regressors, such as non_steady_state_*, unless the model is produced by the tool that identifies outliers (in this case).

Or we would need to have some fixed set of prefixes that are shorthand for a collection, such as cosine expands to the equivalent of cosine_*.