CompCor issues

[chrisgorgo]

I am having issues with CompCor after this was merged #1968

First mask_file input of aCompCor does not work:

            Traceback (most recent call last):
  File "/usr/local/miniconda/lib/python3.6/site-packages/nipype/pipeline/plugins/multiproc.py", line 52, in run_node
    result['result'] = node.run(updatehash=updatehash)
  File "/usr/local/miniconda/lib/python3.6/site-packages/nipype/pipeline/engine/nodes.py", line 366, in run
    self._run_interface()
  File "/usr/local/miniconda/lib/python3.6/site-packages/nipype/pipeline/engine/nodes.py", line 476, in _run_interface
    self._result = self._run_command(execute)
  File "/usr/local/miniconda/lib/python3.6/site-packages/nipype/pipeline/engine/nodes.py", line 607, in _run_command
    result = self._interface.run()
  File "/usr/local/miniconda/lib/python3.6/site-packages/nipype/interfaces/base.py", line 1085, in run
    runtime = self._run_wrapper(runtime)
  File "/usr/local/miniconda/lib/python3.6/site-packages/nipype/interfaces/base.py", line 1033, in _run_wrapper
    runtime = self._run_interface(runtime)
  File "/usr/local/miniconda/lib/python3.6/site-packages/niworkflows/common/report.py", line 52, in _run_interface
    ReportCapableInterface, self)._run_interface(runtime)
  File "/usr/local/miniconda/lib/python3.6/site-packages/nipype/algorithms/confounds.py", line 378, in _run_interface
    self.inputs.mask_files = self.inputs.mask_files[0]
TypeError: '_Undefined' object does not support indexing
Interface ACompCorRPT failed to run. 

Secondly the high_variance_mask returned by tCompCor seems wrong
Before the PR was merged:

After the PR was merged:

This is the call to tcompcor in fmriprep https://github.com/poldracklab/fmriprep/blob/5e3ba02b515ba07c598fda7f4ad7b2bbbf7a8735/fmriprep/workflows/confounds.py#L180 As you can see it should be restricted by highly eroded brain mask which was the case before that PR was merged.

[satra]

@chrisfilo - it's possible that both are because the api has changed. you should now set mask_files and not mask_file.

@mgxd - the deprecation warning will only come online after we release 0.13.0. and i should have checked to see an on_trait_change that sets the other trait by default.

since these interfaces were not released as part of 0.12.0, i would now suggest removing the mask_file trait altogether.

[satra]

@chrisfilo - with regard to previous comment and removal of mask_file completely let us know. what @mgxd implemented in fact makes the interface more useful when one wants to use multiple masks.

[chrisgorgo]

The second problem was observed after switching to the new names: mask_files and high_variance_masks.

[satra]

@chrisfilo - sorry taking a closer look.

[chrisgorgo]

I had a closer look. It turns out I got confused in my testing - the wrong results happen only if mask_file input is set. If the new mask_files input is used the returned high_variance_mask is correct.

I am ok with mask_file being removed instead of trying to fix it.

[mgxd]

@chrisfilo good to hear!

[chrisgorgo]

I wasn't the only one who run into this: https://neurostars.org/t/nipype-tcompcor-error/415

[mgxd]

it seems the MapNode bug is happening because of an interaction between the super() call from TCompCor to CompCor and the node's outputs field being populated with the defaults. One workaround would be to add the defaults within the _run_interface method, but that seems crude.

ipdb> self.outputs
Bunch(components_file='components_file.txt', header=<undefined>, high_variance_masks=<undefined>, ignore_exception=False, mask_files=<undefined>, mask_index=<undefined>, merge_method=<undefined>, num_components=6, realigned_file=<undefined>, regress_poly_degree=1, use_regress_poly=True)

@chrisfilo @satra any ideas?

[satra]

hold on there are several issues with that interface that i am now fixing.