A critical comparison of topology-based pathway analysis methods

Area of Science:

• Bioinformatics
• Computational Biology
• Systems Biology

Background:

• High-throughput gene/protein profiling aims to identify biological processes altered between conditions.
• Pathway analysis methods computationally identify these altered processes.
• Current methods include enrichment-based and topology-based approaches, with limited comparative assessment of the latter.

Purpose of the Study:

• To critically assess the performance of seven topology-based pathway analysis methods.
• To evaluate method sensitivity to experimental factors like sample size and gene filtering.
• To provide recommendations for selecting appropriate pathway analysis tools.

Main Methods:

• Assessed seven representative topology-based pathway analysis methods: SPIA, PRS, CePa, TAPPA, TopologyGSA, Clipper, and DEGraph.
• Conducted controlled experiments evaluating sensitivity to sample size, pathway size, gene filtering, and preprocessing strategies.
• Verified type I error rates and assessed the impact of gene/pathway alterations on detection.

Main Results:

• Demonstrated wide variability in the performance of the tested pathway analysis methods.
• Identified sensitivities to sample size, pathway size, and gene filtering thresholds.
• Showcased differential ability to exploit pathway topological information.

Conclusions:

• The performance of topology-based pathway analysis methods varies significantly.
• Method selection should consider experimental design and data characteristics.
• Recommendations are provided to guide the informed choice of pathway analysis tools for gene expression data.