MNE Python

MNE-Python is a Python package for analysing electrophysiology (MEG, EEG, sEEG, ECoG, NIRS, etc) data.

MNE-Python Versions

Bear Apps has several versions of MNE-Python as modules.

Bear Modules

The following bash loads mne version 1.3.1 and its dependencies - an equivalent is availiable for JupyterLab.

module load bear-apps/2022a
module load MNE-Python/1.3.1-foss-2022a

Note that MNE-Python depends on a number of other python applications that will be loaded automatically. The above code will also load numpy, scipy, numba, matplotlib, sklearn and many other packages that are needed by MNE into your environment.

import mne
import numpy as np

MNE in a virtual environment

If you want to use a specific version of MNE-Python that isn't supported by BEAR, you can install it into a virtual environment. This bash script provides an example. We load Python 3.9.5, create an environment and install MNE into the environment using pip.

#!/bin/bash

module purge;
module load bear-apps/2022a
module load MNE-Python/1.3.1-foss-2022a
module load IPython/7.25.0-GCCcore-10.3.0

export VENV_DIR="${HOME}/virtual-environments"
export VENV_PATH="${VENV_DIR}/mne-example-${BB_CPU}"

# Create master dir if necessary
mkdir -p ${VENV_DIR}
echo ${VENV_PATH}

# Check if virtual environment exists and create it if not
if [[ ! -d ${VENV_PATH} ]]; then
    python3 -m venv --system-site-packages ${VENV_PATH}
fi

# Activate virtual environment
source ${VENV_PATH}/bin/activate

# Any additional installations
pip install mne==1.1.0

As with other examples, this can be copied directly into the terminal or saved as an shell script that can be executed in a terminal.

Evoked response example

The following code is adapted from the MNE-Python overvew of MEG/EEG analysis tutorial. It will download a small example file and run a quick analysis.

You can run this code directly in a Python session on BEAR within an activated MNE environment. The file will be saved into your RDS home directory (eg /rds/homes/q/quinna) unless specified otherwise - please change the sample_data_folder variable if you'd like to save this file elsewhere.

import numpy as np
import mne

# This will download example data into your RDS home directory by default -
# change the next line if you want to save the file elsewhere!
sample_data_folder = '/rds/homes/q/quinna/mne-data/'
sample_data_raw_file = (sample_data_folder / 'MEG' / 'sample' /
                        'sample_audvis_filt-0-40_raw.fif')
raw = mne.io.read_raw_fif(sample_data_raw_file)

events = mne.find_events(raw, stim_channel='STI 014')

event_dict = {'auditory/left': 1, 'auditory/right': 2, 'visual/left': 3,
              'visual/right': 4, 'smiley': 5, 'buttonpress': 32}

reject_criteria = dict(mag=4000e-15,     # 4000 fT
                       grad=4000e-13,    # 4000 fT/cm
                       eeg=150e-6,       # 150 µV
                       eog=250e-6)       # 250 µV

epochs = mne.Epochs(raw, events, event_id=event_dict, tmin=-0.2, tmax=0.5,
                    reject=reject_criteria, preload=True)

conds_we_care_about = ['auditory/left', 'auditory/right',
                       'visual/left', 'visual/right']
epochs.equalize_event_counts(conds_we_care_about)  # this operates in-place
aud_epochs = epochs['auditory']
vis_epochs = epochs['visual']

vis_evoked = vis_epochs.average()

fig = vis_evoked.plot_joint()
fig[0].savefig('my-mne-evoked-example-grad.png')
fig[1].savefig('my-mne-evoked-example-mag.png')
fig[2].savefig('my-mne-evoked-example-eeg.png')

We can save this as 'mne_python_example.py` as our core analysis script. This can be execute from a terminal session in which the appropriate python environment has been loaded. For example, we could open a 'BlueBEAR GUI' session, open a new terminal, change directory to the location of our script and run the following code:

module load bear-apps/2022a
module load MNE-Python/1.3.1-foss-2022a

python mne_python_example.py

At the end you should have some new figures created next to your script.

MNE on the cluster example

Warning

There is currently a bug with this example to do with the automatic file downloading when running on the cluster. Running a simliar example on your own data fromo RDS should work fine.

Alternatively, we can run our evoked responses analysis on the cluster. For this we'll need a job submission script. Let's use the following shell code to load the MNE module from BEAR and run our code.

#!/bin/bash
#SBATCH --account quinna-example-project
#SBATCH --qos bbdefault

module purge; module load bluebear

module load bear-apps/2022a
module load MNE-Python/1.3.1-foss-2022a

python mne_python_example.py

Note

We could equally create or load a virtual environment with a customised Python set-up in our job script. Here we use the built in BEAR module as it is all we need for the case in hand.

We can save the job script as run_mne_python_example.sh and submit it to the cluster using sbatch

sbatch run_mne_python_example.sh

You can see the progress of the job using the 'Active Jobs' page on BEAR portal or by reading the log files.