.png){kind=icon}
.png){kind=icon}
Return
Rényi eigen-entropy: A data-based method for perturbation discrimination in gene regulatory networks
.png){kind=icon}
PRE
AI
Z
R
DOI：10.1016/j.physa.2026.131787.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Different perturbations can induce identical phenotypic outcomes, a phenomenon analogous to the observation that distinct pharmacological interventions may produce equivalent therapeutic effects. However, discriminating between perturbations that lead to qualitatively equivalent steady states remains a fundamental challenge in both biological and mathematical systems. Traditional analytical methods, e.g., bifurcation analysis and sensitivity analysis, either fail to effectively discriminate between distinct perturbations exhibiting similar dynamic impacts, or necessitate reliance on a priori mathematical models. These limitations significantly constrain their applicability in data-driven research scenarios. To address this methodological challenge, we propose a novel framework termed Rényi eigen-entropy (REE). Built upon the recently proposed eigen-entropy (EE) metric, this approach integrates the Rényi order parameter α, thereby allowing the metric to reweight weak versus dominant spectral modes and to discriminate subtle perturbation-dependent differences more flexibly. REE facilitates the discrimination of distinct perturbation routes by analyzing steady-state and transient pre-bifurcation trajectories. We validate the REE method on two systems: a reduced three-node regulatory circuit that simulates early T cell development and the epithelial–mesenchymal transition (EMT) network. In both cases, bifurcation analysis reveals that different perturbations result in qualitatively equivalent steady states. The REE method successfully distinguishes the corresponding pre-bifurcation trajectories. Compared with the EE, REE demonstrates more significant differences under different perturbations, thereby exhibiting superior discriminability. This advantage persists even in the presence of transient dynamics and additive noise. Additionally, we introduce an aggregate separation metric, denoted as AURC Δ, which serves as a compact auxiliary metric to measure the perturbation-route discriminability across the tested range of Rényi orders.
Journal
P
IF:
3.1
Papers:
1.3K
Citations:
3.6W
Organization
Cited Papers