Sharing Research Code

Just as Open Access broadens access to research publications, Open Source Software extends openness to the tools and methods used in research. Open-source software is freely available to use, modify, and redistribute, with source code openly shared so that anyone can inspect how it works, adapt it to specific needs, and contribute improvements. This approach directly supports the core values of Open Science by fostering transparency, reproducibility, and collaboration.

For researchers, open-source tools offer major practical advantages. They reduce dependence on costly proprietary software and are often developed and maintained by active communities that continuously improve functionality based on real research needs. Their flexibility makes them particularly well suited to scientific contexts where methods evolve rapidly and innovation is essential.

In this context, code, scripts, notebooks, and software developed during a research project are scientific outputs in their own right. They document how data are collected, processed, analysed, or visualised, and they play a key role in ensuring that results can be understood and reproduced. Within Horizon Europe, sharing research software is encouraged in the same way as sharing datasets, whenever this is legally and technically possible.

Research code can take many forms, ranging from simple scripts (e.g. a Python script to clean data, an R function for visualisation, a Jupyter notebook with analyses) to complex processing pipelines or full software packages developed in languages such as MATLAB, Java, or C++. Regardless of its complexity, code should be considered for dissemination from the outset of the project.

Before sharing, code should be reasonably well structured and accompanied by basic documentation. This typically includes a README file explaining the purpose of the code, how to run it, required dependencies, and links to associated datasets, as well as a clear statement of the licence. Version control is particularly important in research, as it documents the evolution of analyses, preserves the integrity of results, and makes it possible to return to earlier versions if needed.

For long-term preservation and stable citation, it is recommended to link code repositories to trusted archiving platforms such as Software Heritage or HAL, which assign a persistent identifier and allow the software to be cited like a publication. A recognised open-source licence — such as MIT, Apache 2.0, CeCILL, or GPL—should be chosen depending on the desired level of reuse and compatibility with existing dependencies.

Sharing code does not mean that it must be perfectly optimised or ready for industrial use. What matters is that others can understand the approach, reproduce the results, and potentially reuse or adapt the tools. In this way, sharing research software increases the scientific value, visibility, and impact of the work, and can lead to new collaborations, citations, and constructive feedback. As with datasets, code can be made public alongside the publication it supports or at the time of thesis submission, with any constraints (e.g. patents, confidentiality, security issues) clearly documented in the Data Management Plan.

📎Open licences for software, scripts and code

Scripts and software fall under a different legal framework than data. They must be covered by a free and open-source software licence. The most common options include:

• MIT: a very permissive licence that authorises any reuse, modification and distribution, including commercial, provided that the author is credited.
• GPL v3: allows reuse, but requires that any derivative software must also be distributed under the GPL licence (the “copyleft” principle).
• Apache 2.0: similar to MIT, but with specific provisions for patent management.
• CeCILL: a French open-source licence, compatible with European standards, often used in public research institutions in France.

For your scripts, notebooks or data processing code, a MIT or Apache 2.0 licence is generally sufficient and fully compatible with dissemination on GitHub or Zenodo.

Additional Resources :

Self-Assessment Quiz