A Bayesian framework for genome-wide circadian rhythmicity biomarker detection

Area of Science:

• Chronobiology
• Genomics
• Biostatistics

Background:

• Circadian rhythms, driven by gene feedback loops, regulate essential physiological and behavioral processes.
• Accurate detection of circadian biomarkers is vital, especially with expanding omics datasets.
• Existing methods for circadian rhythm analysis in transcriptomics face limitations.

Purpose of the Study:

• To develop a novel Bayesian framework for detecting circadian rhythms in genome-wide transcriptomic data.
• To incorporate prior biological knowledge and control for multiple testing using a false discovery rate (FDR) approach.
• To provide a robust and accurate analytical tool for circadian gene identification.

Main Methods:

• Developed a Bayesian hierarchical model incorporating prior biological information.
• Utilized a reverse jump Markov chain Monte Carlo (RJMCMC) technique for model selection.
• Implemented a false discovery rate (FDR) approach for multiple testing adjustment.

Main Results:

• The proposed method, BayesCircRhy, demonstrated superior FDR control compared to competing methods in simulations.
• BayesCircRhy showed robustness against various error distributions and outperformed existing approaches.
• Validated in human and mouse RNA-sequencing data, identifying known and novel circadian genes.

Conclusions:

• BayesCircRhy offers a powerful and reliable framework for circadian rhythm detection in transcriptomics.
• The method enhances the identification of circadian genes, contributing to a better understanding of health and aging.
• This Bayesian approach provides a valuable tool for analyzing complex biological rhythms in omics studies.