neurodatascience / nipoppy-nimhans_YLO

Fork for YLO PD setup at nimhans

nipoppy

A workflow manager for curating MRI and tabular data and standardized processing.

Pull-Organize-Process-Push-Yey!

Note: This is a nimhans-YLO dataset-specifc fork

Documentation

nipoppy modules

• nipoppy

Individual containerized pipelines:

• Heudionv
• MRIQC
• fMRIPrep
• TractoFlow
• MAGeT Brain

---

Quickstart (How to run a preset workflow setup)

Onetime setup

1. Project
   • Create a project dir on your local machine: mkdir /home/<user>/projects/<my_project>
   • Create containers, code, data dirs inside your project dir.
2. Containers (Singulaity)
   • Install Singularity
   • Download containers (e.g. Heudiconv) for the pipelines used in this workflow inside the containers dir. For example:
     • cd /home/<user>/projects/<my_project>/containers
     • singularity pull docker://nipy/heudiconv:0.13.1
     • singularity pull docker://nipreps/fmriprep:23.1.3
3. Code
   • Change dir to code: cd /home/<user>/projects/<my_project>/code/
   • Create a new venv: python3 -m venv nipoppy_env
   • Activate your env: source nipoppy_env/bin/activate
   • Clone this repo: git clone https://github.com/neurodatascience/nipoppy-nimhans_YLO.git
   • Install python dependencies: pip install -e .
4. Data
   • Create nipoppy dataset-tree: tree.py /home/<user>/projects/<my_project>/data/<study_name>
   • Create and populate <study_name>/proc/global_configs.json
   • Copy your participant-level dicom dirs (e.g. MNI001, MNI002 ...) into <study_name>/scratch/raw_dicom/
5. RedCap: configure RedCap access. The RedCap query will be used to generate manifest and "bagels" for tabular data dashboard.
   • Create and save the query in <study_name>/proc/.redcap.json
     • See sample redcap.json: workflow/tabular/sample_redcap.json

Periodic runs

1. Change dir to nipoppy code: cd /home/<user>/projects/<my_project>/code/nipoppy
2. Activate your env: source nipoppy/bin/activate (if not already active)
3. Run nipoppy: python nipoppy.py --global_config <> --session_id <> --n_jobs <>
   • This will run "workflows" listed inside the global_configs.json. Currently we are supporting "generate_manifest", "dicom_org", "bids_conv" workflows.
   • The "generate_manifest" workflow will update the current list of participants and also generate bagels from the RedCap report.
4. Dashboard: Upload the bagel i.e. <study_name>/tabular/bagel.csv file to the Dashboard to visualize and query current clinical data.
5. Run processing pipelines (e.g. mriqc, fmriprep, tractoflow): Currently this is setup to run either a single subject locally or batch-jobs on a cluster. Support for local batch-jobs would be added soon.

Expected output

1. <study_name>/dicom: Participant-level dirs with symlinks to the dicom files in the raw_dicom dir
   • Note: dicoms that are unreadable or contain derived (i.e. scanner processed) scans will be skipped and enlisted in the <study_name>/scratch/logs
2. <study_name>/bids: BIDS dataset comprising all the modalities in Nifti format (i.e. nii.gz and sidecar json)
3. <study_name>/derivatives/<proc_pipe>: Output from processing pipelines.

---