Error when setting analyze_channels using mne bids pipeline

💬 Support & Discussions
, ,
1

Hi everyone,

I am using the mne bids pipeline to process my EEG (currently on a single subject for testing). I first ran the pipeline on all channels, which successfully completed. I then restricted the channels to analyze for evoked potentials to a set of posterior-occip electrodes (I am interested in MMN).

MNE version: 1.6
MNE bids pipeline version: 1.5.0
OS: Ubuntu 22.04.2 LTS

# analyze_channels: Union[
#     Literal["all"], Literal["ch_types"], Iterable["str"]
# ] = "ch_types"
analyze_channels = ['PO7','PO8','POz','PO4','PO3','O1','Oz','O2']

┌────────┬ sensor/_01_make_evoked ─────────────────────────────────────────────────────────────────────────────────────────────────────────
│11:36:13│ ⏳️ sub-p015 ses-eeg Input: sub-p015_ses-eeg_task-seqPred_proc-clean_epo.fif
│11:36:13│ ⏳️ sub-p015 ses-eeg Output: sub-p015_ses-eeg_task-seqPred_ave.fif
│11:36:13│ ⏳️ sub-p015 ses-eeg Creating evoked data based on experimental conditions …
│11:36:14│ ⏳️ sub-p015 ses-eeg Contrasting evoked responses …
│11:36:14│ ⏳️ sub-p015 ses-eeg Adding 2 evoked responses and 1 contrast to the report.
│11:36:14│ ⏳️ sub-p015 ses-eeg Adding config and sys info to report
│11:36:16│ ⏳️ sub-p015 ses-eeg Saving report: /RAID1/jupytertmp/ahechler/data/eeg/derivatives/mne-bids-pipeline/sub-p015/ses-eeg/eeg/sub-p015_ses-eeg_task-seqPred_report.html
│11:36:17│ ❌ sub-p015 ses-eeg A critical error occurred. The error message was: 'types.SimpleNamespace' object has no attribute 'eeg_reference'

Aborting pipeline run. The traceback is:

  File "/home/tumnic/ahechler/.conda/envs/pr_andre_2024/envs/mne/lib/python3.12/site-packages/mne_bids_pipeline/_run.py", line 54, in __mne_bids_pipeline_failsafe_wrapper__
    out = memory.cache(func)(*args, **kwargs)
          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/tumnic/ahechler/.conda/envs/pr_andre_2024/envs/mne/lib/python3.12/site-packages/mne_bids_pipeline/_run.py", line 267, in wrapper
    memorized_func(*args, **kwargs)
  File "/home/tumnic/ahechler/.conda/envs/pr_andre_2024/envs/mne/lib/python3.12/site-packages/joblib/memory.py", line 655, in __call__
    return self._cached_call(args, kwargs)[0]
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/tumnic/ahechler/.conda/envs/pr_andre_2024/envs/mne/lib/python3.12/site-packages/joblib/memory.py", line 598, in _cached_call
    out, metadata = self.call(*args, **kwargs)
                    ^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/tumnic/ahechler/.conda/envs/pr_andre_2024/envs/mne/lib/python3.12/site-packages/joblib/memory.py", line 856, in call
    output = self.func(*args, **kwargs)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/tumnic/ahechler/.conda/envs/pr_andre_2024/envs/mne/lib/python3.12/site-packages/mne_bids_pipeline/steps/sensor/_01_make_evoked.py", line 131, in run_evoked
    _restrict_analyze_channels(evoked, cfg)
  File "/home/tumnic/ahechler/.conda/envs/pr_andre_2024/envs/mne/lib/python3.12/site-packages/mne_bids_pipeline/_config_utils.py", line 419, in _restrict_analyze_channels
    elif "eeg" in cfg.ch_types and cfg.eeg_reference == "average":
                                   ^^^^^^^^^^^^^^^^^

It seems like the reference electrodes were never correctly set. This is weird because the pipeline worked before I restricted the channels.

# eeg_reference: Union[Literal["average"], str, Iterable["str"]] = "average"
eeg_reference=['TP9','TP10']

I also checked that the channels are correctly named by comparing against the _electrodes.tsv created by BIDS converting my BrainVision files. Which other reasons could there be for this error?

Thanks for the help!

André

2

Hello and welcome to the forum! This does indeed seem like a bug (and one that would be easy to fix!). I may have time to look into the problem this weekend.

cc @larsoner

Best wishes,
Richard

3

Great, thanks for checking in so quickly! Let me know if you need more info about my dataset. I meanwhile replicated the error with another subject and also when changing the reference.

4

This still seems to be an issue in mne_bids_pipeline 1.6.0.

1 Like
5

Hello, I believe I found the issue and I’m proposing a fix here:

6

We have released MNE-BIDS-Pipeline 1.7, which fixes this issue.

Richard

7

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.