Finding Clocks in Genes: A Bayesian Approach to Estimate Periodicity

Area of Science:

• Genomics
• Systems Biology
• Computational Biology

Background:

• Rhythmic gene expression is vital for understanding biological processes like metabolic cycles and circadian rhythms.
• Existing algorithms like JTK_CYCLE and ARSER have limitations in handling low temporal resolution data or detecting multiple rhythmic categories simultaneously.

Purpose of the Study:

• To introduce a novel algorithm, autoregressive Bayesian spectral regression (ABSR), for estimating gene expression periodicity and classifying multiple rhythmic categories.
• To evaluate ABSR's performance against existing methods, particularly for time-course data with low temporal resolution.

Main Methods:

• Development of the autoregressive Bayesian spectral regression (ABSR) algorithm.
• Simulation studies comparing ABSR with JTK_CYCLE and ARSER.
• Application of ABSR to existing mouse liver time-course data.

Main Results:

• ABSR significantly enhances the accuracy of periodicity estimation and rhythmic category clustering compared to JTK_CYCLE and ARSER for low temporal resolution data.
• ABSR demonstrates robustness and is insensitive to various rhythmic patterns.
• Analysis of mouse liver data revealed ABSR's capability to classify genes into ultradian, circadian, and arrhythmic categories, detecting a substantial overlap with high-resolution methods.

Conclusions:

• ABSR offers a powerful and accurate approach for analyzing rhythmic gene expression, especially when high temporal resolution data is unavailable.
• The algorithm facilitates simultaneous identification of multiple rhythmic categories, providing deeper insights into gene regulatory networks.
• ABSR's application to real-world data validates its effectiveness in classifying gene expression rhythms, including circadian patterns, with reduced sampling frequency.