MNE Python Tutorial 2021

From MEG Core
Jump to navigation Jump to search

Steps to prepare prior to tutorial

Prepare python environment - For 3D rendering and interactive plots, this should be installed on your local computer

 !!IF YOU DO NOT HAVE MINICONDA/ANACONDA INSTALLED -  have IT install miniconda under your user account!!
 !!If you already have an mne environment, you can use another name for the environment and adjust accordingly!!
 
 conda activate base 
 conda install -n base mamba -c conda-forge -y  
 mamba create -n mne conda-forge::mne main::pip main:jupyter -y
 conda activate mne

Download the sample data and scripts.

The data is available from the resources below (Approximately 1Gb compressed and 1.8Gb uncompressed). A final jupyter notebook may be sent over email prior to the tutorial - this will replace the V1 version.

NIMH Users: Download here

 scp <USERNAME>@helix.nih.gov:/data/NIMH_scratch/mne_tutorial/mne_tutorial.tar.gz ./
 tar -xvf mne_tutorial.tar.gz

Non-NIMH MEG Users: Download from MEG Data server

 The data is located on the MEG data server and is accessible by all users with accounts
 An email will be sent with the path to the data

Start Jupyter Notebook

 cd mne_tutorial
 jupyter notebook "MNE Python Tutorial-v1.ipynb"

Prep Non-tutorial Data

MRI Prep

Freesurfer reconstruction

 # Run Freesurfer on Dataset (takes 8+ hrs): 
 # Does not run on BRIK/HEAD - convert to .nii first   -- e.g (module load afni; 3dAFNItoNIFTI  anat+orig. )
 
 # (OPTION1) - Install and process freesurfer locally on computer https://surfer.nmr.mgh.harvard.edu/fswiki/rel7downloads
 recon-all -all -i <MRI Used w/ MEG> -s <SUBJECT>
 
 # (OPTION2) - INSTRUCTIONS FOR BIOWULF - Run line by line on the bash terminal
 subjid=       #Set Freesurfer ID
 mri=          #Set MRI Name - must be a nifti file NOT BRIK/HEAD
 export SUBJECTS_DIR=      #Set output folder, Make sure this directory exists
 module load freesurfer
 echo -e '#!/bin/bash\nrecon-all -all  -i ' ${mri} -s ${subjid} | sbatch  --mem=3g --time=24:00:00     #SUBMITS JOB TO SBATCH

Watershed Processing for Boundary Element Model

 #In bash terminal (5-10minutes)
 #Requires freesurfer to be installed
 subjid=<SUBJECT>
 subjects_dir=<SUBJECTS_DIR>
 conda activate mne
 python -c "import mne; mne.bem.make_watershed_bem(subject='${subjid}', subjects_dir='${subjects_dir}')"

MEG Prep

 # The events of interest must be denoted in the MarkerFile.mrk
 https://megcore.nih.gov/index.php?title=Pyctf
 # The marker file can also be processed with the hv_proc code that is distributed with the data
 # A limited demonstration on the day of the tutorial will be given to process the data triggers into the MarkerFile.

Copy Data To Tutorial Folder

 cp -R ${SUBJECTS_DIR}/${subjid} mne_tutorial/SUBJECTS_DIR/    
 #Move the subject's freesurfer reconstruction to the mne_tutorial folder.  If needed rename the output freesurfer subject folder to the 8 digit meg id found in the meg data
 
 cp -R <MEGDATASET.ds> mne_tutorial/
 cp {BRIK/HEAD w/ fiducials marked OR Brainsight .txt file } mne_tutorial/${SUBJID}