A comparative study of topology-based pathway enrichment analysis methods.

Area of Science:

• Bioinformatics
• Systems Biology
• Computational Biology

Background:

• Pathway enrichment analysis is crucial for interpreting Omics data, revealing functional roles of biomolecules.
• Existing methods primarily use expression levels, while newer approaches incorporate pathway topology for improved accuracy.
• A systematic comparison of these network-based methods is lacking, hindering optimal tool selection.

Purpose of the Study:

• To systematically evaluate and compare the performance of nine network-based pathway enrichment analysis methods.
• To assess method performance across diverse real-world Omics datasets with varying feature and sample sizes.
• To identify the most suitable methods for different biological contexts and data types.

Main Methods:

• Comparative analysis of nine network-based pathway enrichment methods.
• Utilized three real Omics datasets with varying numbers of features and samples.
• Evaluated methods based on their ability to detect pathway enrichment using expression and/or network topology information.

Main Results:

• Significant methodological and empirical differences were observed among the nine evaluated methods.
• Methods integrating both expression levels and pathway topology demonstrated superior statistical power, particularly for metabolomics data.
• NetGSA, a method considering both differential expression and pathway topology, showed enhanced performance.

Conclusions:

• For genomic data with large pathways, several methods perform comparably well.
• NetGSA exhibits superior performance for small pathways, common in metabolomics, by integrating expression and topological changes.
• The study highlights the advantage of topology-aware methods for specific biological data types and pathway sizes.