Abstract : The complexity of analysis pipelines in biomedical sciences poses a severe challenge for the transparency and reproducibility of results. Researchers are increasingly incorporating software development technologies and methods into their analyses, but this is a quickly evolving landscape and teams may lack the capabilities to set up their own complex IT infrastructure to aid reproducibility. Basing a reproducible research strategy on readily available solutions with zero or low set-up costs whilst maintaining technological flexibility to incorporate domain-specific software tools is therefore of key importance. We outline a practical approach for robust reproducibility of analysis results. In our examples, we rely exclusively on established open-source tools and free services. Special emphasis is put on the integration of these tools with best practices from software development and free online services for the biostatistics domain.
https://hal.inria.fr/hal-03100421
Contributor : Boris Hejblum <>
Submitted on : Thursday, January 7, 2021 - 7:27:51 PM Last modification on : Monday, January 11, 2021 - 12:21:01 PM
Boris Hejblum, Kevin Kunzmann, Ennio Lavagnini, Anna Hutchinson, David Robertson, et al.. Realistic and Robust Reproducible Research for Biostatistics. 2021. ⟨hal-03100421⟩