Quantifying periodicity in omics data

Area of Science:

• * Biological rhythms and oscillations
• * Computational biology
• * Signal processing

Background:

• * Biological systems exhibit oscillations across various timescales (ultradian, circadian, etc.).
• * Existing methods for periodicity detection often assume sinusoidal models and fail to capture waveform complexity or multiple periodicities.
• * Accurate analysis of biological oscillations is crucial for understanding underlying biological dynamics.

Purpose of the Study:

• * To develop a novel Fourier-based computational method for analyzing biological oscillations.
• * To quantify waveform shape and detect multiple periodicities in oscillatory biological data.
• * To provide a robust tool for analyzing noisy datasets and high-throughput biological analyses.

Main Methods:

• * Development of a Fourier-based measure to generate a de-noised waveform from significant frequencies.
• * Correlation of the generated waveform with raw data (e.g., yeast respiratory oscillations).
• * Calculation of oscillation statistics, including waveform metrics and multi-periodicity.

Main Results:

• * The novel method successfully generates de-noised waveforms capturing complex oscillatory patterns.
• * The approach quantifies essential oscillation statistics, including waveform shape and multiple periods.
• * The method demonstrates utility in analyzing noisy datasets and high-throughput biological data like metabolomics and flow cytometry.

Conclusions:

• * The developed Fourier-based method offers a significant advancement over traditional techniques for analyzing biological oscillations.
• * This approach provides deeper insights into biological dynamics by characterizing waveform shape and multi-periodicity.
• * The tool is applicable to a wide range of noisy and high-throughput biological datasets, enhancing discovery potential.