Workflows and Provenance: Toward Information Science Solutions for the Natural Sciences

Area of Science:

• Computational Science
• Biomolecular Research
• Data Science

Background:

• Ubiquitous computation and big data present challenges for research reproducibility.
• Computational methods may seem self-documenting but require explicit documentation for reproducibility.
• Reproducibility in computation mirrors laboratory experiment requirements.

Purpose of the Study:

• To address concerns about ensuring reproducibility in computational research environments.
• To highlight the necessity of documenting computational workflows and execution details.
• To discuss computational reproducibility in the context of biomolecular Nuclear Magnetic Resonance spectroscopy (bioNMR) and the PRIMAD model.

Main Methods:

• Discussing the documentation of protocols (workflows) in computational research.
• Detailing the necessary components of the computational environment for reproducibility: algorithms, implementations, software, data, and execution logs.
• Examining these aspects within the framework of biomolecular Nuclear Magnetic Resonance spectroscopy (bioNMR).

Main Results:

• Computational reproducibility necessitates comprehensive documentation of both workflows and execution environments.
• Specific application to biomolecular Nuclear Magnetic Resonance spectroscopy (bioNMR) demonstrates these principles.
• The PRIMAD model is presented as a framework for achieving computational reproducibility.

Conclusions:

• Reproducibility in computational research is achievable through meticulous documentation of processes and environments.
• Detailed logging and workflow management are essential for validating computational results.
• The study emphasizes the importance of standardized documentation for big data and computational science.