Mne bids pipeline

Skip Background

Use on Biowulf

Intro

BIDS is a standard specification for neuroimaging/physiology data. This currently includes at least: MRI, fMRI, DTI, EEG, MEG, fNIRS (and possibly ECOG/sEEG). BIDS typically describes how RAW data is organized - and processed data is located in the bids_dir/derivatives/{AnalysisPackage}/{SUBJECT}/... The main advantage is that common code can be generated to process data organized in a standard format. Therefore, you should be able to import the bids data into any number of neurophysiological packages (MNE, Brainstorm, SPM, Fieldtrip, ...). Additionally, standardized processing packages known as BIDS apps can be used to process the data in the same way as long as the data is organized in BIDS.

 BIDS (formal specs) >> MNE_BIDS (python neurophysiological BIDS) >> MNE_BIDS_PIPELINE (structured processing in MNE python according to BIDS definition)

BIDS website

 https://bids-specification.readthedocs.io/en/stable/

Bids data organization

MNE Bids

MNE bids is a python tool to make and access neurophysiological data. This tool is a branch of the MNE tools for neurophysiological analysis.

MNE Bids website

 https://mne.tools/mne-bids/stable/index.html

MNE bids pipeline background

Mne bids pipeline website:

 https://mne.tools/mne-bids-pipeline/

Full example:

 https://mne.tools/mne-bids-pipeline/examples/ds000248.html

Processing

All processing is defined in the config.py file. This has hundreds of options to define the processing.

 Here are some of the typical options:  https://mne.tools/mne-bids-pipeline/settings/general.html

Here is the full list of options:

 https://github.com/mne-tools/mne-bids-pipeline/blob/main/config.py

Example Config

(copy into a config.py text file and use)

Modify for use with your data - at a minimum: bids_root, conditions

 study_name = 'TESTSTudy'
 bids_root = 'YOUR_BIDS_DIR' #<< Modify
 task = 'TASKNAME'   #If this is left empty, it will likely find another task and throw an error
 l_freq = 1.0
 h_freq = 100.
 epochs_tmin = -0.1
 epochs_tmax = 0.2
 baseline = (-0.1, 0.0)
 resample_sfreq = 300.0
 ch_types = ['meg']
 conditions = ['stim']  #<< list of conditions
 N_JOBS=4  #On biowulf - leave this empty - it will set N_JOBS equal to the cpu cores assigned in sinteractive
 on_error = 'continue'  # This is often helpful when doing multiple subjects.  If 1 subject fails processing stops
 
 #crop_runs = [0, 900] #Can be useful for long tasks that are ended early but full file is written.

Use on biowulf

!!If you run into any issues - please let Jeff Stout know so that the code can be updated!!

Make MEG modules accessible

#Run the line below every login 
#Or add the following line to your ${HOME}/.bashrc (you will need to re-login or type bash to initialize)
module use --append /data/MEGmodules/modulefiles

Start interactive session with scratch to render visualization offscreen

 sinteractive --mem=6G --cpus-per-task=4 --gres=lscratch:50  #adjust mem and cpus accordingly - subjects will run in parrallel

Create BIDS data from MEG data

 module load mne_scripts
 
 #Afni coregistered mri data  
 make_meg_bids.py -meg_input_dir ${MEGFOLDER} -mri_brik ${AFNI_COREGED}+orig.BRIK
 
       ## OR ##
 
 #Brainsight coregistered mri data
 make_meg_bids.py -meg_input_dir ${MEGFOLDER} -mri_bsight ${MRI_BSIGHT}.nii -mri_bsight_elec ${EXPORTED_BRAINSIGHT}.txt

Freesurfer Processing

If you have already processed your freesurfer data - just copy or link the freesurfer data(see below). You can also set the freesurfer directory in mne-bids-pipeline config file. Alternatively, you can have the pipieline calculate freesurfer as part of the processing. This will take time - adjust sinteractive session (see below).

Already Processed - Copy Data

 mkdir -p ${bids_dir}/derivatives/freesurfer/subjects  #If this folder doesn't already exists
 #Copy your subject from the freesurfer directory to the bids DERIVATIVES freesurfer directory
 cp -R ${SUBJECTS_DIR}/${SUBJID}  ${bids_dir}/derivatives/freesurfer/subjects

Process Using MNE-BIDS-Pipeline

 #your sinteractive session must include --time=24:00:00  so that the freesurfer processing doesn't time out
 
 #add the following step to your mne_bids_pipeline processing
 --steps=freesurfer

Process data using MNE Bids Pipeline

First make a config file: example

 module purge 
 module load mne_bids_pipeline
 mne-bids-pipeline-run.py --config=CONFIG.py
       #Optional Flags
       #--steps=preprocessing,sensor,source,report  or all (default) - Can be a list of steps or a single step
       #--subject=SUBJECTID(without the sub- prefix)

Example Output Images (from anonymized data - the pipeline does not anonymize)

TEST DATA for biowulf

This section provides some test data to analyze using mne-bids-pipeline. Feel free to adjust parameters in the config.py after you run through the analysis the first time. All of the analysis parameters are defined in the config.py provided. Additional config.py parameters can be found above in Processing.

Start Interactive Session

 sinteractive --mem=6G --cpus-per-task=4 --gres=lscratch:50

Copy and untar the data into your folder

 cp -R /vf/users/MEGmodules/modules/bids_example_data_airpuff.tar ./
 tar -xvf bids_example_data_airpuff.tar
 
 #Add the bids_root to your config file
 echo bids_root=\'$(pwd)/bids_example_data_airpuff\' >> $(pwd)/bids_example_data_airpuff/config.py

Load module and process the data

 module load mne_bids_pipeline
 mne-bids-pipeline-run.py --config=$(pwd)/bids_example_data_airpuff/config.py

 #Copy this path for the next step
 echo $(pwd)/bids_example_data_airpuff/derivatives/mne-bids-pipeline/sub-ON02747/ses-01/meg/sub-ON02747_ses-01_task-airpuff_report.html

 #In another terminal download your results from biowulf
 scp ${USERNAME}@helix.nih.gov:${PathFromAbove} ./

 Open sub-ON02747_ses-01_task-airpuff_report.html in an internet browser to view the report