From Data to Code

what are Data clusters?

Recherche Data Gouv Ecosystem

local competence centers dedicated to supporting researchers in FAIR policy implementation
different profiles (Librarians, Archivists, Research Engineers, Data Protection Officer)
different research institutions operating on the same territory
collaborate with other Data clusters on specific topics (work group 5 : research source code)
national Repository is central in a Data Cluster’s activity

map of data clusters on mainland France

Data clusters are part of the global ecosystem framed by the National Data Repository ‘Recherche Data Gouv’ and its catalogue (still in beta version). Data is central in our activity but to some extent, because data are processed by software and we need both of them to reproduce experiences, source code is also within the scope of the data clusters ; there is an ongoing debate to know up to which point source code should be covered by data curators, since both objects are very different : data are static objects, after the publication of the results, data won’t change much, while source code is more dynamic and may involve other developers or even communities long after the publications are made. Besides, some say that skills needed for data curation are not the same as skills required by source code maintenance, promotion and reproducibility

ARDoISE Datahub

committee : 16 members + corresponding experts

centralizes all matters related to the code through guichet-ardoise
helps researchers to write their DMP
trains them to FAIR data and dissemination of research data and source code
helps them to use data repositories, especialy Research Data Gouv
advises them on data preservation and safety
advises them on legal and ethical aspects of data management
Provides them with information on source code dissemination and reproducibility
guides them in the use of lab notebooks
receives them in dedicated places in University Rennes 2’s library

ARDoISE data cluster logo

Sometimes, I feel like…

University policies in Open Source Software

working in pair, but with different and multiple connections

Damien Belvèze

member of Working Group 5 within Research Data Gouv ecosystem
webinars with INRIA members
tests Guix package manager

Kilian Heutte

experienced in R and Rmarkdown
works with MSHB research engineers
tests data anonymisation tools

What services do Data Clusters provide to research community regarding research source code?

communicate on Mastodon / LinkedIn on source code sharing and preserving best practices
showcase source code linked to projects they followed up
write pages on their websites or blogposts on hypotheses wordpress websites on source code management
webinars : how to describe source code and software in a research project (example here)

📓 Bourdaret et al. (2026)

other stakeholders than Data Clusters take part to source code tracking and promotion:

groups of research engineers focused on source code management
University Libraries (cf. University of Nanterre’s Library)
OSPOs (see University of Grenoble’s OSPO)
ARDoISE Data Cluster advocates for becoming part of the future Rennes academic OSPO

“University OSPOs are the local execution engines for Open Source Software policies” 📓 Di Cosmo (2025)

(figure) 📓 Young et al. (2024)

Integration in OSPOs

Academic Open Source Programme Offices are built by federating university services concerned with Open Source Programmes :

Intellectual property office, Technology Transfer Office
research teams and labs
libraries (involved in curating HAL and Recherche Data Gouv portals)
computing centers and experimentation centers
stakeholders from the local open source industry (including non profit organisations such as OW2)

services adapted to different targets

target	Experience	project’s type and dimension
🎓 PhD Students	are learning to code	scripts
💡 development for individual needs made by researchers	trained users	scripts, add-ons to known software
💡 🔧 collaborative development	state of the art development	comprehensive software

🎓 💡 🔧 for all

🎓 Support for PhD Students

basic programming skills (R / Python) acquired during their master

documentation (codemeta, SMP)
development good practices (virtual env)
archiving and citing : Software Heritage Archive, SWHID
promotion of open source programmes to process research data

🎓 PhD training

data visualisation with Quarto (introduction to Quarto with R and Rstudio)
initiation to R and Rmarkdown
introduction to Jupyter notebooks
initiation to Git and Gitlab
Data Anonymisation tools
how to open, share and preserve reproducible source code

💡 individual researchers

they often have already acquired strong development practices since their PhD by collaborating with research engineers

Questions about licences
archiving and citing : Software Heritage Archive, SWHID

💡 🔧 research teams

discussions with software engineers about :

visibility : HAL, catalogue-esr.gouv.fr
archiving and citing : Software Heritage Archive, SWHID

Looking for the good level:

Data Cluster? (but more focused on research data)
Library? (focused on linking publications to underlying source code)
Library / Data Cluster as member of an OSPO? (what will be our role in this new frame)
Other?

References

memes are made with framameme generator

Bourdaret, C., Leclerc, C., Ronat, M., & Sepulveda, B. (2026). Codes et logiciels pour la recherche (p. 50) [Gestion de Projet]. ENSSIB.

Di Cosmo, R. (2025). Academic OSPOs in France - supporting the software pillar of open science.

Young, Barba, L., Choudhury, S., Flannagan, Lippert, D., & Littauer, R. (2024). A definition of an Academic OSPO.