Noise Characteristics of Molecular Oscillations in Simple Genetic Oscillatory Systems

We study the noise characteristics of stochastic oscillations in protein number dynamics of simple genetic oscillatory systems. Using the three-component negative feedback transcription regulatory system called the repressilator as a prototypical example, we quantify the degree of fluctuations in oscillation periods and amplitudes, as well as the noise propagation along the regulatory cascade in the stable oscillation regime via dynamic Monte Carlo simulations. For the single protein-species level, the fluctuation in the oscillation amplitudes is found to be larger than that of the oscillation periods, the distributions of which are reasonably described by the Weibull distribution and the Gaussian tail, respectively. Correlations between successive periods and between successive amplitudes, respectively, are measured to assess the noise propagation properties, which are found to decay faster for the amplitude than for the period. The local fluctuation property is also studied.

💡 Research Summary

This paper investigates the stochastic properties of molecular oscillations in a minimal genetic circuit, using the three‑component negative‑feedback loop known as the repressilator as a prototypical example. The authors adopt a fully stochastic simulation framework—dynamic Monte Carlo (Gillespie) algorithm—to capture the intrinsic noise arising from transcription, translation, and degradation events at the single‑cell level. By scanning parameter regimes that support stable limit‑cycle behavior, they generate long time series of protein copy numbers for each of the three genes and extract the periods and amplitudes of successive oscillations.

Statistical analysis of the collected data reveals distinct distributional characteristics for the two observables. Amplitude fluctuations follow a Weibull distribution with a pronounced right‑hand tail, indicating that large excursions are rare but more probable than would be expected under a symmetric distribution. In contrast, the period distribution is approximately Gaussian near its mean, with a Gaussian tail for unusually long periods, suggesting that the timing of the oscillation is relatively tightly regulated but still subject to occasional large delays. The authors quantify the relative magnitude of these fluctuations, finding that amplitude variability exceeds period variability under the same conditions.

To assess how noise propagates through the cascade, the paper computes autocorrelation functions for successive periods and successive amplitudes. Amplitude autocorrelations decay rapidly: the correlation between adjacent cycles is already modest, and by the second lag the correlation is essentially zero. Period autocorrelations, however, persist over several cycles, decaying more slowly and retaining a measurable positive value up to three or four periods. This asymmetry implies that the circuit retains memory of its timing more robustly than of its magnitude, a feature that could be advantageous for biological clocks that need to keep a reliable rhythm while allowing amplitude to adapt to fluctuating cellular conditions.

The study also examines “local” noise at each stage of gene expression. Translational steps contribute the largest variance, whereas transcriptional noise is comparatively modest, and degradation adds a smaller but non‑negligible component. This hierarchy aligns with theoretical expectations that protein synthesis is the dominant source of stochasticity in gene‑regulatory networks.

In the discussion, the authors connect these quantitative findings to broader biological implications. The rapid decay of amplitude noise suggests that the repressilator can quickly dampen spurious fluctuations, preserving the fidelity of downstream processes that may be sensitive to protein concentration. The slower decay of period noise indicates that the timing information is more resilient, supporting synchronization with external cues or other cellular oscillators. By characterizing the full probability distributions and correlation structures, the work provides a statistical foundation for designing synthetic genetic clocks with desired noise properties and for interpreting experimental single‑cell time‑course data.

Overall, the paper demonstrates that even a simple three‑gene negative‑feedback loop exhibits rich stochastic behavior: non‑Gaussian amplitude distributions, near‑Gaussian period distributions, distinct noise propagation dynamics, and stage‑specific contributions to variability. These insights deepen our understanding of how intrinsic molecular noise shapes the performance of genetic oscillators and offer practical guidelines for engineering robust synthetic circuits.