Investigating reproducibility and tracking provenance - A genomic workflow case study

Area of Science:

• Computational biology
• Genomics
• Bioinformatics

Background:

• Computational bioinformatics workflows are crucial for genomics data analysis.
• Ensuring workflow reproducibility is a significant challenge due to incomplete understanding of requirements and assumptions.
• Tracking provenance information is vital for capturing requirements and supporting workflow reusability.

Purpose of the Study:

• To identify implicit assumptions in common workflow definition and execution approaches.
• To propose recommendations for mitigating these assumptions and improving reproducibility.
• To guide the scientific community towards achieving reproducible computational genomic analyses.

Main Methods:

• Implementation of a complex bioinformatics workflow using three distinct approaches.
• Analysis of workflow definition and execution to identify implicit assumptions.
• Development of a set of recommendations based on identified challenges.

Main Results:

• Implicit assumptions in workflow approaches lead to insufficient documentation and execution failures.
• The study identified specific challenges hindering reproducibility in bioinformatics.
• Recommendations are proposed to address these challenges and enhance workflow reliability.

Conclusions:

• Reproducing, adapting, or repeating bioinformatics workflows demands significant technical expertise and adherence to reproducibility requirements.
• Explicit workflow specification and the proposed recommendations are key to enhancing the reproducibility of computational genomic analyses.
• Addressing implicit assumptions is crucial for overcoming the reproducibility crisis in computational science.