2026-10-28 –, LUMC02
Do you struggle to reproduce analyses months later or reuse someone else’s workflow? This session shows you how to make your research executable, reusable, and FAIR-compliant using persistent Jupyter notebook environments. You will discover how the ICOS Carbon Portal provides access to persistently identified datasets while supporting complete analytical workflows.
Jupyter notebooks combine code, documentation, and results, but rapidly changing software dependencies often prevent reliable reproduction. You will see how capturing notebooks, code, and dependencies in archived Docker images allows you to preserve the full execution environment and rerun analyses on demand.
By the end, you will understand how to publish executable workflows with persistent identifiers, preserve computational environments, and reduce reproducibility challenges—helping you keep your research accessible and trustworthy over time.
Authors: Claudio D’Onofrio, Ute Karstens, Zois Zogopoulos, Alex Vermeulen, Carbon Portal Team
Do you find that your analyses stop working after a few months—or that reproducing someone else’s results feels impossible? In this session, you will explore a practical approach to making your research workflows reproducible, reusable, and aligned with FAIR principles.
You will discover how the ICOS Carbon Portal supports researchers with collaborative tools and direct access to persistently identified datasets, ensuring your data remains findable and consistently referenced throughout your work. But reproducibility is not just about data—you also need to preserve the full analytical process.
Jupyter notebooks are widely used because they combine executable code with explanations, figures, and results in a single document. This makes them ideal for documenting scientific workflows. However, you have likely experienced how quickly software environments change. Even if a notebook is preserved and assigned a persistent identifier, differences in dependencies can make it impossible to rerun and verify results later.
In this session, you will learn how to overcome this challenge by capturing entire computational environments. You will see how Docker images can bundle your notebooks, source code, and all required dependencies into a portable, archived unit. These environments can then be reactivated on demand via the ICOS Jupyter Hub, allowing you—or others—to rerun analyses exactly as originally executed.
After this session, you will:
• Understand how to publish executable Jupyter notebooks with persistent identifiers and rich metadata
• Learn how to preserve and share complete computational environments using Docker
• Gain practical insights into reducing reproducibility issues caused by evolving software dependencies
While no solution can fully eliminate all challenges—such as reliance on external data sources—this approach significantly lowers the barrier to sharing reproducible research. You will leave with concrete ideas you can apply to make your own workflows more transparent, reliable, and ready for long-term reuse.
