Installation

Installation#

MEGPrep is officially distributed as a container image. The containerized workflow is recommended because it provides the most reproducible runtime and avoids most local dependency conflicts.

If Docker cannot be installed, the Docker daemon is unavailable, or the image cannot be pulled in your network environment, use the local development installation workflow. The local workflow installs MEGPrep from source and can run without a Docker image, but local system libraries and package versions may affect reproducibility.

Manual Docker Installation#

Install Docker according to your operating system. For detailed installation instructions, visit the Docker official website.

Check Docker:

docker info

Pull the MEGPrep image:

docker pull cmrlab/megprep:<version>

Replace <version> with a release tag such as 0.0.3 or latest.

Alternative: Local Installation Without Docker#

The scripts under scripts/install-dev/ provide a source-based local installation path for Linux environments where container installation is not available or image pulling is blocked. This workflow installs or reuses Conda, Nextflow, FreeSurfer, and MEGPrep source dependencies in a local installation directory.

bash scripts/install-dev/install_megprep_dev_linux.sh
bash scripts/install-dev/install_megprep_dev_linux.sh --install-dir /data/megprep-dev
bash scripts/install-dev/install_megprep_dev_linux.sh --no-freesurfer

After installation, load the generated environment:

source <install-dir>/env.sh

See scripts/install-dev/README.md for the full local installation workflow.

Docker Entry Point Options#

Use --steps as the primary way to choose the pipeline stage. For example, use --steps anatomy for structural MRI only and --steps meg_all for the full MEG workflow with existing anatomy.

docker run --rm -it cmrlab/megprep:<version> -h

Docker runs do not require Docker’s --user flag. The container entrypoint prepares mounted output permissions, then drops to the host UID/GID inferred from /input before running Nextflow. Report-only runs that only mount /output infer ownership from /output. If that inference is not appropriate, pass -e LOCAL_UID="$(id -u)" -e LOCAL_GID="$(id -g)" to choose the output owner explicitly.

Usage: /program/nextflow/run.sh [options]
Options:
  -c, --config          Specify the Nextflow config file
  -i, --input           Specify the input directory
  -o, --output          Specify the output directory including report results
  -s, --steps           Pipeline mode, for example all, meg_all, anatomy, report
  -r, --view-report     Run Streamlit to view the report and do not run Nextflow
  --static_task_log_mode failed|all-command-log|none
  --fs_license_file     Specify the FreeSurfer license file
  --fs_subjects_dir     Specify the FreeSurfer SUBJECTS_DIR directory
  --t1_dir              Specify the T1 image directory
  --t1_input_type       Specify the T1 input type
  --t1_dicom_series_glob Optional relative glob for selecting DICOM series
  --resume              Resume the previous run

Common --steps values are meg_all for full MEG processing with existing anatomy, all for anatomy plus full MEG, anatomy for structural MRI only, and report for static report regeneration. See Configuration Reference for all modes and modifiers. --static_task_log_mode controls whether the static report bundles only failed task logs, successful .command.log files as well, or no command logs.

Contents