Beyond Linear Additive and Hill Functions: A General Logistic Reformulation of Delay-Coupled Gene Regulatory Networks with Equilibrium Analysis, Hopf Bifurcation, and Lipschitz Stability

Ismail Belgacem · 2026

Read Beyond Linear Additive and Hill Functions: A General Logistic Reformulation of Delay-Coupled Gene Regulatory Networks with Equilibrium Analysis, Hopf Bifurcation, and Lipschitz Stability by Ismail Belgacem (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-26810.txt.

Introduction

Belgacem replaces Hill functions, the standard sigmoidal curves in gene-network models, with logistic-based models that avoid three flaws. Hill curves can blow up, turn complex-valued, or trap a network in its off state when fitted at non-integer powers. His logistic forms stay smooth, real-valued, and positive at zero input while keeping the familiar S-shape. On a two-gene network with delayed mutual activation and self-repression, he derives matching rules, locates the single feasible steady state, and maps where stability gives way to oscillation.

Catalog record

CREATORIsmail Belgacem
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE80138
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2604.26810

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Beyond Linear Additive and Hill Functions: A General Logistic Reformulation of Delay-Coupled Gene Regulatory Networks with Equilibrium Analysis, Hopf Bifurcation, and Lipschitz Stability Ismail Belgacem Abstract Hill functions, the dominant tool for modeling gene regulatory networks, carry fundamental limitations: when the cooperativity exponent takes a non-integer value, as it routinely does when fitting experimental dose-response data, derivatives diverge at the origin, complex-valued arithmetic silently corrupts ODE trajectories, and the zero output at zero activation concentration traps models irreversibly in off-states that no intrinsic dynamics can escape. This paper uses logistic functions based models that are globally C ∞ C^{\infty} , real-valued for all arguments, and strictly positive at zero concentration. The substitution resolves all three pathologies simultaneously while preserving the sigmoidal dynamics that make Hill functions biologically appealing. Working with the delay-coupled two-gene mutual-activation and self-repression network of Vinoth et al. as a concrete model system, we analyze two logistic reformulations of the Hill-based delay differential equations: a linear additive activation combined with logistic self-repression, and a fully sigmoidal formulation in which both activation and self-repression are governed by logistic functions. A closed-form parameter matching relationship λ = n / θ \lambda=n/\theta is used by equating the slopes of logistic and Hill functions at their half-maximal points, ensuring local input–output equivalence. Closed-form parameter expressions for the weighted logistic formulation, κ 1 = 4 ​ g A \kappa_{1}=4g_{A} , θ B = g A \theta_{B}=g_{A} , and λ 1 = ln ⁡ 3 / g A \lambda_{1}=\ln 3/g_{A} , are derived by simultaneously matching basal expression rates and local activation slopes to the original Hill-linear hybrid model. For the linear additive logistic formulation, the unique biologically feasible equilibrium is computed at ( 167.96 , 164.22 ) (167.96,\,164.22) nM; for the weighted logistic formulation it is computed at ( 144.46 , 139.99 ) (144.46,\,139.99) nM, the reduction arising from saturation of the bounded activation term. In the delay-free case ( τ = 0 \tau=0 ), local asymptotic stability is established for both formulations by proving that the Jacobian trace is strictly negative for all positive parameters; the system then remains stable for delays τ ∈ ( 0 , τ c ) \tau\in(0,\tau_{c}) and loses stability via Hopf bifurcation at the critical delay τ c \tau_{c} . Hopf bifurcation analysis, via the full two-equation transcendental system solved numerically, yields critical self-repression delays of τ c ≈ 1.19 \tau_{c}\approx 1.19 min for the linear additive formulation and τ c ≈ 1.64 \tau_{c}\approx 1.64 min for the weighted logistic formulation, with higher-order bifurcation sequences characterised numerically in each case; direct numerical simulation of complex dynamics and extreme events in the logistic reformulations is identified as an important direction for future work. A quantitative comparison of the two formulations shows that replacing linear additive activation with weighted logistic activation reduces the global Lipschitz constant L F L_{F} from 18.74 18.74 min -1 to 5.54 5.54 min -1 , an approximately 70% reduction permitting substantially larger integration steps, while the Jacobian Lipschitz constant L D ​ F L_{DF} decreases from 0.24 0.24 to 0.084 0.084 nM -1 min -1 . 1 Introduction Gene regulatory networks constitute the fundamental control architecture of living cells, orchestrating the spatiotemporal patterns of gene expression that underlie development, homeostasis, and adaptation to environmental change. Mathematical modeling of these networks has become an indispensable tool for understanding the dynamical behaviors that emerge from transcriptional interactions — oscillations, bistability, multistability, and chaos [ 1 , 2 ] — and for guiding the rational design of synthetic gene circuits in biotechnology and medicine. At the heart of nearly every such model sits the Hill function, a formulation introduced over a century ago to describe cooperative ligand binding and subsequently adopted as the near-universal representation of sigmoidal gene regulatory responses. Its two forms, the activation function h^{+}(x,\theta,n)=x^{n}/(x^{n}+\theta^{n}) and the repression counterpart h^{-}(x,\theta,n)=\theta^{n}/(x^{n}+\theta^{n}) , are intuitive, mechanistically grounded, and have been applied successfully to models of the lambda phage lysis–lysogeny decision, the lac and gal operons in Escherichia coli , developmental patterning in Drosophila , and mammalian cell cycle control, among many others. Complementary mathematical work has rigorously analyzed transcription-translation systems and enzymatic reaction networks [ 3 , 4 , 5 , 6 , 7 , 8 , 9 ] , and has explored chaos, bifurcations, and hybrid dynamics in ring-circuit and piecewise-linear gene network architectures [ 1 , 2 , 10 ] . The Hill coefficient n quantifies cooperativity and \theta the half-maximal concentration, both carrying clear physical interpretations, and the sigmoidal shape faithfully captures the switch-like regulatory transitions characteristic of biological decision-making. Yet this success conceals substantial hidden costs. The activation Hill function vanishes identically at zero input, h^{+}(0,\theta,n)=0 . This is mathematically convenient but biologically unrealistic. Experimental studies consistently show minimal basal transcription (0.1–10% of maximal) due to stochastic promoter binding, incomplete repressor occupancy, and chromatin dynamics. Models built on Hill functions either ignore this or introduce ad hoc additive offsets. A second set of limitations concerns mathematical structure and numerical behavior. When the Hill coefficient takes a non-integer value — as it routinely does when fitting experimental dose-response curves, yielding values such as n\approx 1.39 , 2.73 , or 3.52 — the function loses global smoothness. For n\in(k,k+1) , derivatives of order greater than k diverge at the origin, restricting the function to C^{k} smoothness and rendering unavailable the higher-order analytical tools — center manifold reduction, normal form analysis, Hessian-based optimization — that modern bifurcation and parameter estimation workflows depend upon. The power-law form x^{n} is also computationally expensive for non-integer exponents, since it requires evaluating the transcendental composition e^{n\ln x} , which accumulates floating-point error near zero and, critically, becomes complex-valued whenever any trajectory component transiently overshoots to a negative value, as every adaptive ODE solver will produce from rounding errors alone. The result is silent corruption: the solver continues integrating a complex-valued surrogate system, producing smooth and visually plausible trajectories that are not solutions to the true biological model. Beyond smoothness, the maximum slope of h^{+} at its inflection point equals n/(4\theta) , coupling threshold position and transition steepness into a single expression. Threshold and steepness cannot be adjusted independently, which obscures biological parameter interpretation and complicates the rational tuning of synthetic circuits where engineers need to set sensitivity and decision threshold as separate design variables. Finally, the rational form x^{n}/(x^{n}+\theta^{n}) has no closed-form inverse for general n , blocking the feedback linearization and gradient-based model predictive control strategies [ 11 , 12 ] that are increasingly central to experimental implementation, while the zero production rate at x=0 creates fundamental controllability gaps from which Hill-based dynamics cannot recover without external intervention. The logistic function resolves all of these limitations simultaneously. Its activation form f^{+}(x,\theta,\lambda)=1/(1+e^{-\lambda(x-\theta)}) and repression form f^{-}(x,\theta,\lambda)=1/(1+e^{\lambda(x-\theta)}) are globally C^{\infty} , real-valued for all arguments including negative ones, and strictly positive at zero concentration for all finite \lambda and \theta . The elegant self-referential derivative f^{\prime}=\lambda f(1-f) reduces Jacobian entries to products of function values, eliminating fractional exponents from stability analysis entirely. The closed-form logit inverse f^{-1}(y)=\theta+\lambda^{-1}\ln(y/(1-y)) enables exact feedback linearization. The parameters \theta and \lambda are fully decoupled: the threshold can be repositioned without altering the transition slope, and vice versa, and both map directly to biologically measurable quantities

Read the complete text