Order restricted inference for oscillatory systems for detecting rhythmic signals.

Area of Science:

• Genomics
• Systems Biology
• Bioinformatics

Background:

• Biological processes like cell cycles and circadian rhythms are governed by complex oscillatory systems.
• Identifying rhythmic components, such as circadian genes, from time-course gene expression data is challenging.
• Existing methods risk misclassifying rhythmic and non-rhythmic genes.

Purpose of the Study:

• To develop a robust methodology for detecting rhythmic signals in biological systems.
• To introduce Order Restricted Inference for Oscillatory Systems (ORIOS) as a novel approach.
• To overcome limitations of model-based methods for rhythmic gene identification.

Main Methods:

• Developed Order Restricted Inference for Oscillatory Systems (ORIOS), a constrained inference methodology.
• ORIOS utilizes mathematical inequalities rather than predefined mathematical functions (e.g., sinusoidal) to define rhythmicity.
• Evaluated ORIOS performance using simulated data and real-world datasets from mouse tissues and cell lines.

Main Results:

• ORIOS demonstrated substantially higher power in detecting true rhythmic genes compared to popular existing methods.
• The ORIOS methodology significantly reduced the misclassification of non-rhythmic genes as rhythmic.
• Results were consistent across diverse gene expression patterns and biological samples.

Conclusions:

• ORIOS provides a robust and powerful tool for identifying rhythmic biological signals, particularly rhythmic genes.
• The inequality-based approach makes ORIOS adaptable and not constrained by specific mathematical models.
• This method improves the accuracy of rhythmic gene identification in time-course expression data.