We aim to develop an analysis framework to facilitate reproducible and flexible neuroimaging analyses and allow the assessment and optimisation of the reproducibility of such analyses.
We also provide some educational materials to emphasize and demonstrate reproducible practices in neuroimaging.
We focus on three main principles to support reproducible analyses.
- Explicit specification of dependency between steps
- Clear recording and visualisation of provenance
- Tight control of loading/unloading tools
- High- and low-level data diagnostics
- Data integrity (checksum) recorded and checked at every step
- Compatible with both MATLAB and OCTAVE
- Tested on both Windows and Linux (Ubuntu)
- Integration of a large variety of MATLAB/OCTAVE-, Linux- and Python-based neuroimaging tools
- Parallel execution on a single workstation and an HPC
- Modular design
- Interface to download data from common sources, including OpenNEURO
- Installation scripts for tools
- Reproducible Analysis (reproa) is the main pipeline system for this project.
- Workshops repository contains various educational materials.