Reproducibility and Provenance

GFRN coordinates national reproducibility networks including the Ouvrir la Science-aligned French network (RFRR), advocating for open and rigorous research practices across disciplines. The Turing Way is a community-led open handbook covering reproducible research, project design, collaboration, and research ethics, serving as a cross-cutting methodological reference across the open science ecosystem. OSF provides the pre-registration and registered reports infrastructure through which researchers commit study designs and analysis plans before data collection, reducing publication bias. Registered Reports, adopted by over 300 journals as of 2022, extends this model further: peer review and provisional acceptance occur before data collection begins, making acceptance conditional on study design quality rather than outcome. arXiv, founded in 1991, established the preprint model that enabled immediate open sharing of scientific work prior to peer review, a practice now central to open science and adopted across biology and medicine through bioRxiv and medRxiv. The quantitative biology section of arXiv (q-bio.NC) remains the primary preprint venue for theoretical and computational neuroscience. Protocols.io is the primary platform for sharing versioned, citable step-by-step research protocols, addressing the methods reporting gap that contributes to irreproducibility in published research. The technical infrastructure for computational reproducibility centres on ReproNim, which develops NIDM as its primary provenance standard, built formally on PROV-O, and integrates DataLad for distributed dataset versioning and re-executable analysis workflows. BrainLife.io implements this stack in a cloud environment, applying automatic provenance tracking to neuroimaging pipelines run against BIDS-organised data from OpenNeuro and DANDI Archive. Together, PROV-O, NIDM, and DataLad constitute the provenance layer that makes computational workflows machine-readable and re-executable independent of local infrastructure.