Towards reliable pathway enrichment

Pathway enrichment analysis is crucial for understanding genomic data. But misuse of this method and inaccurate reporting threaten the reliability of research. This study aims to determine how widespread these issues are and evaluate their impact. Our work will propose solutions to improve best practice. As a result we hope to speed up the process of using research findings to benefit society.

Objective

Improved methodology will have several benefits:

• biomedical researchers will take reproducibility more seriously when it comes to computational methods
• avoiding methodological pitfalls will reduce the failure rate of research translation, helping new therapies materialise sooner
• improved quality will enhance the community's perception of investing in biomedical research.

Timeline

2024–ongoing.

Approach

We use a comprehensive multi-pronged approach:

• we screen published journal articles for the presence of statistical errors and missing methodological data
• for selected high impact articles, we will reproduce their work using valid methods, and report any studies whose conclusions were invalidated
• using bioinformatics investigations, we quantify how these statistical errors distort research findings
• using a survey questionnaire of biomedical researchers, we will investigate root causes of the issue
• we are initiating a consortium of experts to contribute towards a community led set of guidelines for conducting and reporting enrichment analysis
• we are publishing “gold standard” methods, enabling researchers to avoid pitfalls
• if necessary, we will develop new computational software/web resources if we have reason to believe it will help researchers avoid these pitfalls
• tangentially, we are investigating whether web-based bioinformatics tools are in violation of NHMRC/NIH research data preservation guidelines.

Community impact

The research does not directly impact communities. However, improved methodology in this area will have future benefits:

• if successful, biomedical researchers will take reproducibility more seriously when it comes to their computational methods.
• reduced research waste by avoiding methodological pitfalls.
• avoiding such pitfalls reduces the failure rate of research translation, helping new therapies materialize sooner.
• improving quality will enhance community perception of investing in biomedical research.