Dual gene set enrichment analysis (dualGSEA); an R function that enables more robust biological discovery and...

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Gene set enrichment analysis (GSEA) is crucial for interpreting gene expression studies.
• Previous comparisons focused on statistical performance, neglecting downstream biological implications of different GSEA methods.
• The distinction between pairwise and single-sample GSEA approaches requires further investigation.

Purpose of the Study:

• To compare statistical and biological outcomes of pairwise GSEA with various pre-ranking methods.
• To evaluate GSEA, single sample GSEA (ssGSEA), and gene set variation analysis (GSVA) independently.
• To introduce dualGSEA, a novel tool for robust interpretation of GSEA results.

Main Methods:

• Comparative analysis of statistical and biological results from pairwise GSEA and fGSEA using diverse gene pre-ranking strategies.
• Independent assessment of GSEA, ssGSEA, and GSVA.
• Development and implementation of the dualGSEA tool for enhanced result interpretation.

Main Results:

• Pairwise GSEA and fGSEA yielded comparable results across different gene pre-ranking methods.
• Pairwise GSEA demonstrated a tendency to overgeneralize biological enrichment.
• Single-sample analyses revealed a lack of biological distinction between groups identified as statistically significant by pairwise GSEA.
• The dualGSEA tool offers improved interpretation by presenting multiple statistics and visuals.

Conclusions:

• Pairwise GSEA may obscure true biological distinctions by overemphasizing statistical significance.
• Single-sample GSEA approaches are essential for validating biological differences.
• The dualGSEA tool provides a more reliable framework for discovery research by integrating statistical and biological interpretation.
• dualGSEA facilitates clearer differentiation between statistical significance and biological relevance in gene expression data.