Toward practical transparent verifiable and long-term reproducible research using Guix

Area of Science:

• Computational science
• Research methodology
• Open science

Background:

• The scientific community faces a reproducibility crisis, hindering research validation.
• Existing resources facilitate access to methods, data, and code, but neglect computational environments.
• Current methods for describing computational environments (e.g., software versions, container images) lack sufficient detail for true reproducibility.

Purpose of the Study:

• To highlight the critical issue of irreproducible computational environments in scientific research.
• To introduce and demonstrate the utility of open-source tools, specifically Guix, for addressing this challenge.
• To advocate for the adoption of such tools to enhance transparency and reproducibility in scientific computation.

Main Methods:

• Analysis of the limitations in current practices for documenting and sharing computational environments.
• Illustration of how the Guix package manager can be used to define and share reproducible computational environments.
• Demonstration of Guix's capability to manage complex software dependencies for scientific analyses.

Main Results:

• Identified a significant gap in research transparency concerning computational environments.
• Showcased Guix as a viable solution for creating shareable and reproducible scientific computing environments.
• Demonstrated that while not all research steps may be fully reproducible, computational transparency is technically achievable.

Conclusions:

• Computational environment description is crucial for scientific reproducibility and is currently inadequately addressed.
• Open-source tools like Guix offer a practical and effective solution for scientists to ensure the transparency of their computational workflows.
• Widespread adoption of tools promoting computational transparency is essential for advancing open science principles.