- Python 3.8 or higher
- pip or conda package manager
Once meg2bids is published to PyPI, you can install it with:
pip install meg2bids# Clone the repository
git clone https://github.com/ln2t/meg2bids.git
cd meg2bids
# Install in editable mode
pip install -e .
# Or install normally
pip install .# Create a new environment
conda create -n meg2bids python=3.10
conda activate meg2bids
# Clone and install
git clone https://github.com/ln2t/meg2bids.git
cd meg2bids
pip install -e .If you want to contribute to meg2bids:
# Clone the repository
git clone https://github.com/ln2t/meg2bids.git
cd meg2bids
# Install with development dependencies
pip install -e ".[dev]"
# Verify installation
pytest tests/# Check version
python -c "import meg2bids; print(meg2bids.__version__)"
# Run help
python meg2bids.py --helpmeg2bids automatically installs:
mne>=1.0- MEG/EEG analysismne-bids>=0.13- BIDS integrationnumpy>=1.20- Numerical computing
To validate your BIDS output:
# Using npm
npm install -g bids-validator
# Or use npx (no installation needed)
npx bids-validator your_bids_directory/pip install mne mne-bidsOn Linux/macOS, you may need:
pip install --user meg2bidsOr use a virtual environment (recommended):
python -m venv meg2bids-env
source meg2bids-env/bin/activate # On Windows: meg2bids-env\Scripts\activate
pip install meg2bidsCheck your Python version:
python --versionIf you have multiple Python versions, use:
python3.10 -m pip install meg2bids- Read the Tutorial to get started
- Check Configuration Guide for setup details
- See Examples for sample configurations