Neuroimaging article reexecution and reproduction assessment system

Area of Science:

• Computational neuroscience
• Scientific reproducibility
• Data analysis transparency

Background:

• Research article value increasingly depends on complex data analysis.
• Data analysis offers higher transparency and reproducibility potential than data production.
• Fully reexecutable research outputs enable automatic, end-to-end article generation.

Purpose of the Study:

• To develop a robust and portable system for fully reexecutable research articles.
• To address core challenges in full article reexecution.
• To provide a framework for reproducibility assessments.

Main Methods:

• Utilized a peer-reviewed neuroimaging article with existing reexecution instructions as a foundation.
• Designed a modular reexecution system allowing for adaptable code, data, and environment specifications.
• Detailed best practices to mitigate challenges in full article reexecution.

Main Results:

• Developed a modular, robust, and portable system for generating reexecutable research articles.
• Identified and mitigated core challenges associated with full article reexecution.
• Demonstrated the system's capability for reproducibility assessments using statistical metrics and visual inspection.

Conclusions:

• Fully reexecutable articles represent a feasible and valuable best practice in scientific publishing.
• The developed system enhances understanding of data analysis variability and increases trust in research findings.
• Encourages re-use and adaptation of the system for advancing reproducible research.