Configuration Examples

Configuration Examples#

These examples show the parts of nextflow.config that are most commonly changed between studies. Keep the global path settings, container settings, and cluster settings consistent with your local environment.

Full MEG Preprocessing with Existing Anatomy#

Use this when fs_subjects_dir already contains FreeSurfer or DeepPrep reconstructions for the MEG subjects.

params {
    dataset_dir = "/input"
    output_dir = "/output"
    preproc_dir = "${params.output_dir}/preprocessed"
    fs_subjects_dir = "/smri"
    steps = "meg_all"

    dataset_format = "auto"
    meg_import_config = """
    subject_id: null
    session_id: null
    task: null
    run_id: null
    """

    preproc_config = """
    preproc:
      - filter: {l_freq: 0.5, h_freq: 125, method: iir, iir_params: {order: 5, ftype: butter}}
      - notch_filter: {freqs: 50 100}
      - resample: {sfreq: 250}
    """
}

Structural MRI and MEG in One Run#

Use steps = "all" when MEGPrep should run anatomical reconstruction before the full MEG pipeline.

params {
    steps = "all"
    is_bids = true
    anatomy_preprocess_method = "freesurfer"
    t1_dir = "/input"
    fs_subjects_dir = "/smri"
    fs_license = "/fs_license.txt"

    mri_import_config = """
    subject_id: null
    session_id: null
    task: null
    run_id: null
    """
}

Resting-State MEG#

For resting-state recordings, the continuous preprocessing and ICA stages are unchanged. The epoching stage creates fixed-length epochs from the cleaned raw recording.

params {
    steps = "meg_epochs"

    epoch_config = """
    task_type: resting
    resting:
      fixed_length_duration: 2.0
    event_source: find_events
    autoreject: false
    interpolate_bads: false
    drop_bad_channels: false
    exclude_event_id:
      - 255
    epochs:
      event_id: null
      tmin: 0.0
      tmax: 2.0
      reject_by_annotation: true
      picks: meg
      baseline: null
      preload: true
      detrend: null
    """
}

Task-Based MEG with Trigger Channel Events#

Use this when events can be recovered from a stimulus channel with mne.find_events.

params {
    steps = "meg_epochs"

    meg_import_config = """
    subject_id:
      - "01"
      - "02"
    session_id: null
    task:
      - aef
    run_id: null
    """

    epoch_config = """
    task_type: task
    event_source: find_events
    autoreject: false
    find_events:
      stim_channel: null
      shortest_event: 1
      min_duration: 0.0
    epochs:
      event_id: null
      tmin: -0.2
      tmax: 0.8
      reject_by_annotation: true
      picks: meg
      baseline: null
      reject:
        mag: 4e-12
      preload: true
      detrend: null
    """
}

Task-Based MEG with BIDS Events#

Use event_source = "event_file" when trial definitions should come from *_events.tsv files.

params {
    epoch_config = """
    task_type: task
    event_source: event_file
    autoreject: false
    event_file:
      trial_type:
        word_onset_01: 1
        phoneme_onset_01: 2
    epochs:
      event_id: null
      tmin: -0.2
      tmax: 1.0
      reject_by_annotation: true
      picks: meg
      baseline: null
      preload: true
      detrend: null
    """
}

Empty-Room Style Covariance#

MEGPrep currently handles empty-room or noise recordings through covar_type = "raw" and raw_covariance_task_id. The workflow pairs each experimental recording with a raw recording whose filename has the same BIDS entities except for the task label.

params {
    steps = "meg_all"
    covar_type = "raw"
    raw_covariance_task_id = "emptyroom"

    covar_config = """
    compute_raw_covariance:
      tmin: 0
      tmax: null
      method: auto
      reject:
        mag: 4e-12
      reject_by_annotation: true
      rank: info
    """
}

If a dataset contains empty-room files that should not be imported as ordinary experimental recordings in raw discovery mode, exclude them during import and use a BIDS-style or task-specific layout for covariance pairing.

raw_include_keywords:
  - task-rest
raw_exclude_keywords:
  - phantom
  - crosstalk

For CTF raw folders, set file_suffix = ".ds". For BTI/4D folders whose primary data file is named c,rfDC, set file_suffix = "c,rfDC".

Report-Only Run#

Use report to rebuild the static HTML report without re-running MEG or MRI processing.

docker run --rm -it \
  -v /data/bids:/input \
  -v /data/out:/output \
  -v /data/smri:/smri \
  cmrlab/megprep:0.0.3 \
  -i /input -o /output --fs_subjects_dir /smri --steps report

Docker End-to-End Example#

The Docker entrypoint prepares mounted output permissions and then drops to the host UID/GID inferred from /input. The host output directory may be absent when the command starts. Report-only runs that only mount /output infer ownership from /output. If neither mount should determine the output owner, add -e LOCAL_UID="$(id -u)" -e LOCAL_GID="$(id -g)".

docker run --rm -it \
  -v /data/bids:/input \
  -v /data/out:/output \
  -v /data/smri:/smri \
  -v /data/license.txt:/fs_license.txt \
  -v /data/nextflow.config:/program/nextflow/nextflow.config \
  cmrlab/megprep:0.0.3 \
  -i /input \
  -o /output \
  --fs_license_file /fs_license.txt \
  --fs_subjects_dir /smri \
  --steps meg_all \
  --resume

Cluster Example#

On a SLURM cluster, use a Nextflow config profile or process block to select the SLURM executor and container runtime. The important MEGPrep parameters stay the same.

workDir = "/scratch/project/megprep_work"

singularity.enabled = true
singularity.autoMounts = true

process {
    executor = "slurm"
    queue = "general"
    cpus = 4
    memory = "16 GB"
    container = "/containers/megprep_0.0.3.sif"
}

params {
    dataset_dir = "/project/study/bids"
    output_dir = "/project/study/derivatives/megprep"
    fs_subjects_dir = "/project/study/derivatives/smri"
    steps = "meg_all"
}
Contents