Time-Course Gene Set Analysis for Longitudinal Gene Expression Data.

Area of Science:

• Genomics
• Bioinformatics
• Statistical Genetics

Background:

• Gene set analysis (GSA) is effective for cross-sectional genomic data.
• Longitudinal data analysis requires specialized methods for gene expression studies.
• Existing methods may not adequately capture dynamic gene expression patterns over time.

Purpose of the Study:

• To extend gene set analysis to longitudinal data, termed time-course gene set analysis (TcGSA).
• To develop a robust method capable of handling unbalanced longitudinal data with missing measurements.
• To identify gene sets with significant temporal expression variations and compare patterns across biological conditions.

Main Methods:

• Random effects modeling with maximum likelihood estimates.
• Incorporation of all available repeated measurements, accommodating missing at random (MAR) data.
• Hypothesis-driven approach focusing on a priori defined gene sets and their expression heterogeneity over time.

Main Results:

• TcGSA identified 69 significant gene sets in an HIV vaccine trial, a feat missed by standard univariate and GSEA methods.
• The method detected 4 gene sets associated with influenza vaccination in a separate vaccine study.
• Simulation studies confirmed TcGSA's superior statistical properties and increased power for time-course gene set analysis.

Conclusions:

• TcGSA provides a powerful and flexible framework for analyzing time-course gene expression data.
• The method enhances the discovery of biologically relevant temporal patterns in genomic studies.
• TcGSA is available as an R package, facilitating its adoption by the research community.