RepeatFS: a file system providing reproducibility through provenance and automation.

Area of Science:

• Bioinformatics
• Computational Biology
• Data Science

Background:

• Reproducibility is crucial in science, especially with big data.
• Complex bioinformatics pipelines pose challenges for consistent replication and verification.
• Existing frameworks for reproducibility are often cumbersome and require method redesign.

Purpose of the Study:

• To develop a file system that records, replicates, and verifies informatics workflows.
• To address the challenges of reproducibility in big data bioinformatics analysis.
• To promote analytical transparency and simplify the verification of computational methods.

Main Methods:

• Developed RepeatFS, a file system designed for reproducibility.
• RepeatFS records, replicates, and verifies informatics workflows.
• No alteration to original computational methods is required.

Main Results:

• Successfully visualized and replicated diverse bioinformatics tasks with over a million operations.
• RepeatFS accurately identified software inconsistencies leading to replication discrepancies.
• Demonstrated the ability to promote analytical transparency and task automation.

Conclusions:

• RepeatFS effectively enhances reproducibility in bioinformatics.
• The file system simplifies the verification of complex computational workflows.
• RepeatFS offers a non-intrusive solution for ensuring analytical transparency.