Automatise your processing pipelines with nipype / pydra

Title

Automatise your processing pipelines with nipype / pydra

Leaders

David Meunier : Twitter: https://twitter.com/DavidM0579 Mattermost: @david.meunier https://mattermost.brainhack.org/brainhack

Collaborators

To fill at the end

Brainhack Global 2022 Event

Brainhack Marseille

Project Description

Neuroimaging and electrophysiology processing requires many steps, calling different softwares, possibly in different languages (typically, matlab batches or shell scripts).

Nipype has provided an integrative solution, with a sufficient level of complexity to cover most of the needs for writting pipelines in neuroimaging. It is based on the notion workflows, being an orderd succession of nodes, linking inputs and outputs. Nodes can be user-written function (in python), interfaces with existing softwares (e.g. FSL, AFNI or SPM), or even other user-defined sub-workflows.

Nipype is at the base of many widely used docker images, such as fmriprep and qsiprep. And has been extendend for other applications, such as EEG/MEG processing (ephypype), graph analysis in functional connectivity (graphpype) or non-human primate anatomical MRI segmentation (macapype).

Nipype has now achieved a degree of maturity to have become predominant in the community. But some of the limitations still prevails. It has decided in the last years to rewrite the core engine of nipype, to incorporate new functionnalities, such as runnnig containers as one node. The new implementation will be called pydra, and also still in its infancy, we expect it to become a major standard in the community.

Link to project repository/sources

Nipype: https://github.com/nipy/nipype

Pydra: https://github.com/nipype/pydra

Related projects: https://github.com/Macatools/macapype (potentially) https://github.com/neuropycon/ephypype https://github.com/neuropycon/graphpype

Goals for Brainhack Global

In this project, we propose :

• to give an overview of how nipype works
• to advise you if it is useful for your typical processing,
• to help writting specific nodes or workflows dedicated to your processing.

For advaced users, We also propose:

• to explain the advances of pydra compaired to nipype
• to write some tools existing in nipype in pydra.

Good first issues

• write an interface of your own python function
• build your own very simple processing pipelines with 2 nodes (possibly with your onw function) and one connection using pyBIDS inputs
• convert an existing pipeline from nipype to pydra

Communication channels

https://mattermost.brainhack.org/brainhack/channels/bhg22-marseille-auto-nipype-pydra

Skills

Neuroimaging/electrophysiology processing: 100% Shell Script / Matlab Batch: 75% Python: 50% Nipype: 25%

Onboarding documentation

https://macatools.github.io/macapype/contribute.html

What will participants learn?

Writing easily modifiable and reusable pipeline; Contributing to reproducible science

Data to use

None

Number of collaborators

1

Credit to collaborators

Starting a new pydra-based project will be rewarded as a main contributor to the project, and possibly as an author on a subsequent prospective (methological) article. Working on your pipeline will be rewarded as your own github project.

Image

Type

pipeline_development

Development status

0_concept_no_content

Topic

reproducible_scientific_methods

Tools

Nipype

Programming language

Python

Modalities

fMRI

Git skills

2_branches_PRs

Anything else?

Testing is project template is directly usable on the website BrainHack Marseille 2022, Subject to modification.

Things to do after the project is submitted and ready to review.

• Add a comment below the main post of your issue saying: Hi @brainhackorg/project-monitors my project is ready!
• Twitter-sized summary of your project pitch.