Bayesian modeling of factorial time-course data with applications to a bone aging gene expression study

Area of Science:

• Biomedical Sciences
• Genomics
• Bioinformatics

Background:

• Biomedical studies often generate complex data with multiple factors, time points, and units.
• Analyzing simultaneous variations over time and conditions presents significant analytical challenges.
• Understanding gene expression changes during aging is crucial for bone health research.

Purpose of the Study:

• To develop a Bayesian statistical method for analyzing multi-level biological data.
• To detect and characterize signals at factorial, longitudinal, and transcriptional levels.
• To identify gene clusters with similar biological mechanisms in bone aging.

Main Methods:

• Developed a Bayesian statistical model incorporating cluster- and time-point-specific parameters.
• The model determines temporal gene expression profile shapes.
• Applied the methodology to microarray data from mice with varying Bromodomain (Brd2) gene expression.

Main Results:

• Successfully applied the Bayesian method to a bone aging gene expression dataset.
• Identified latent gene clusters based on similar biological mechanisms.
• Characterized transcriptomic changes during bone aging in male and female mice.

Conclusions:

• The developed Bayesian method effectively analyzes complex, multi-level biological data.
• The approach facilitates the discovery of transcriptional networks and gene clusters.
• Provides insights into age-dependent, sex-linked bone loss phenotypes related to Brd2 gene expression.