Extracting the Strongest Signals from Omics Data: Differentially Expressed Pathways and Beyond.

Area of Science:

• Bioinformatics
• Computational Biology
• Systems Biology

Background:

• Gene Set Analysis (GSA) is crucial for interpreting omics data by leveraging biological knowledge.
• GSA facilitates the extraction of robust biological signals compared to analyzing individual genes.
• Existing GSA methods primarily focus on differential expression, limiting comprehensive biological insights.

Purpose of the Study:

• To review and categorize major Gene Set Analysis (GSA) approaches.
• To evaluate GSA methods for differential expression (DE), differential variability (DV), and differential co-expression (DC) between phenotypes.
• To provide guidelines for selecting appropriate GSA methods based on performance.

Main Methods:

• Review of established and novel Gene Set Analysis (GSA) methodologies.
• Comparative power analysis and Type I error rate assessment using simulated data.
• Application and evaluation of DE, DV, and DC GSA approaches on real-world omics datasets.

Main Results:

• Identified three major GSA testing types: differential expression (DE), differential variability (DV), and differential co-expression (DC).
• Comparative analysis revealed varying performance and error rates for different GSA approaches under specific conditions.
• Real-world data application demonstrated that DE, DV, and DC GSA methods yield complementary biological information.

Conclusions:

• The choice of GSA approach significantly impacts the biological insights derived from omics data.
• Differential variability (DV) and differential co-expression (DC) analyses offer valuable information beyond traditional differential expression (DE) testing.
• This study provides a framework for selecting the most effective GSA strategy for specific experimental designs and data types.