Reproduced MEEG Group Study

Code that attempts to reproduce "A Reproducible MEG/EEG Group Study With the MNE Software" by Jas et al., 2018 (link)

To run the analysis:

1. Download the files in this project.
2. Select a folder where you will store the data and set an environment variable reproduction_data to its path.
3. Create a folder bids/derivatives/freesurfer_lk/ and download the contents of the freesurfer_lk component into it.
4. In a terminal, navigate to the repository root.
5. Create and activate the conda environment:

   conda env create -n reproduction -f environment.yml
   conda activate reproduction

6. Run the full analysis with

   snakemake all --cores <n_cores> --keep-going

   Where n_cores is the number of processing cores you have available for the analysis.

Troubleshooting

In case of errors when running Snakemake, look through the output in the terminal to see which rule caused the error.

Rule download_from_openneuro

Sometimes downloading data from openeuro does not work on the first attempt and then we have to try again. There are two workaround:

1. Download the data from the openneuro dataset manually and put it in the bids folder you created earlier. You will need the following files/folders:
   1. All sub-* folder in the root.
   2. The derivatives/meg_derivatives/ folder.
2. Run Snakemake telling it to try again if a job fails. There is currently no way to do this at the rule level, so we will run only the download_from_openneuro rule:

   snakemake --until download_from_openneuro --cores <n_cores> --restart-times 5

   Once done, run the full workflow again (the downloaded files will not be downloaded again).

Rules apply_linear_filter or make_artifact_epochs

These rules run linear filtering at some point which may create issues when another job is running at the same time. In Snakefile, we told Snakemake that if the filtering_process resource is set during Snakemake invocation, then those two rules will require as much of this resource as there are cores available and thus these rules can't be run in parallel with each other job (see resources nodes of these rules). This setting does not change anything unless we run Snakemake with the --resources flag and set filtering_process resource capacity to the total number of cores. To avoid these rules running in parallel with other rules we also set the default amount of this resource to 1 for all the rules that do not have this resource specified with the default-resources flag.

snakemake --cores <n_cores> --resources filtering_process=<n_cores> --default-resources filtering_process=1 --keep-going