Switching, Multiple Time-Scales and Geometric Blow-Up in a Low-Dimensional Gene Regulatory Network
S. Jelbart, K. U. Kristiansen & P. Szmolyan · 2025
Read Switching, Multiple Time-Scales and Geometric Blow-Up in a Low-Dimensional Gene Regulatory Network by S. Jelbart, K. U. Kristiansen & P. Szmolyan (2025) 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-2504-20467.txt.
Introduction
Samuel Jelbart and co-authors untangle two kinds of reduction in a model of a gene regulatory network, in a 2025 paper. Gene network equations often split into fast and slow parts. They also contain steep switches that act almost like steps. Modelers simplify with quasi-steady reductions or piecewise-smooth swaps. Whether those two shortcuts commute was unclear.
A four-dimensional network serves as the test case. Geometric blow-up analysis tracks the dynamics near the tricky points. The interplay of time scales and switching turns out to be delicate. Wrong ordering of reductions can distort the model's behavior. Applied mathematicians gain a worked warning. Systems biologists learn when their simplified models stay faithful.
Catalog record
| CREATOR | S. Jelbart, K. U. Kristiansen & P. Szmolyan |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 93 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2504-20467
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Switching, Multiple Time-Scales and Geometric Blow-Up in a Low-Dimensional Gene Regulatory Network S. Jelbart, K. U. Kristiansen & P. Szmolyan Abstract ODE-based models for gene regulatory networks (GRNs) can often be formulated as smooth singular perturbation problems with multiple small parameters, some of which are related to time-scale separation, whereas others are related to ‘switching’, (proximity to a non-smooth singular limit). This motivates the study of reduced models obtained after (i) quasi-steady state reduction (QSSR), which utilises the time-scale separation, and (ii) piecewise-smooth approximations, which reduce the nonlinearity of the model by viewing highly nonlinear sigmoidal terms as singular perturbations of step functions. We investigate the interplay between the reduction methods (i)-(ii), in the context of a 4-dimensional GRN which has been used as a low-dimensional representative of an important class of (generally high-dimensional) GRN models in the literature. We begin by identifying a region in the small parameter plane for which this problem can be formulated as a smooth singularly perturbed system on a blown-up space, uniformly in the switching parameter. This allows us to apply Fenichel’s coordinate-free theorems and obtain a rigorous reduction to a 2-dimensional system, that is a perturbation of the QSSR. Finally, we show that the reduced system features a Hopf bifurcation which does not appear in the QSSR system, due to the influence of higher order terms. Taken together, our findings suggest that the relative size of the small parameters is important for the validity of QSS reductions and the determination of qualitative dynamics in GRN models more generally. Although the focus is on the 4-dimensional GRN, our approach is applicable to higher dimensions. Keywords: gene regulatory networks, model reduction, geometric singular perturbation theory, geometric blow-up, switching MSC2020: 34E13, 34E15, 34E10, 94C11, 37N25 1 Introduction Mathematical models of biological and biochemical phenomena often feature switching , i.e., abrupt dynamical transitions which occur when a particular variable or set of variables in the model cross over a threshold. In the context of ODE-based models, switching can often be described by smooth singular perturbation problems which limit to piecewise-smooth (PWS) systems as one or more of the perturbation parameters tend to zero. Problems of this kind have received a lot of attention in recent years; here we refer to [ 24 , 27 , 33 , 44 , 48 , 49 ] for recent work on low-dimensional problems featuring switching in ODE models of intracellular calcium oscillation, substrate-depletion oscillation, mitotic oscillation in a model for the embryonic cell cycle, a cAMP signalling model for cell aggregation and the Jansen-Rit model in mathematical neuroscience. In addition to switching, many biological and biochemical processes – and consequently also the mathematical models thereof – evolve on multiple time-scales. This is also true of the particular applications mentioned above, and one of the main contributions of the references [ 24 , 27 , 33 , 44 , 48 ] has been to demonstrate that switching and multiple time-scale aspects of the models can be analysed with a common set of mathematical techniques based on geometric singular perturbation theory (GSPT) [ 11 , 25 , 37 , 53 , 55 ] and a method of desingularisation known as geometric blow-up [ 8 , 30 , 35 , 36 , 50 ] . Singular perturbation analyses of problems involving both switching and multiple time-scales tend to be complicated by the presence of multiple small parameters, some of which relate to switching, whereas others are related to the multi-scale structure of the model. One approach to the analysis of these problems is to group the small parameters together by relating them to a single small parameter, as in e.g. [ 24 , 34 , 48 ] . This serves to simplify the analysis, however, approaches of this kind tend to rely on the availability of good numerical estimates for the parameter values, and one cannot in general expect the same dynamics if the singular limit is approached along a different curve in parameter space. Detailed mathematical analyses of problems with multiple small parameters that describe the asymptotic behaviour and singular limit in the corresponding small parameter space, where the small parameters are kept independent, are much more involved. Perhaps unsurprisingly, analyses of this kind appear to be less common in the literature, however we refer to [ 1 , 5 , 29 , 38 ] for notable exceptions. 1.1 Multiple time-scales and switching in GRNs This article is motivated by the interplay between multiple time-scale dynamics and switching in an important class of gene regulatory network (GRN) models which can be traced back to the pioneering work in [ 17 , 51 ] . A significant body of research investigating the dynamics and properties of these systems has emerged in recent decades, see e.g. [ 10 , 41 , 42 , 46 , 47 ] and the many references therein. The models in question can be written as 2N -dimensional ODE systems of the form \begin{split}r_{i}^{\prime}&=F_{{i}}(h_{{\color[rgb]{0,0,0}{i,1}}}(p_{1};\theta_{1},n_{1}),h_{{\color[rgb]{0,0,0}{i,2}}}(p_{2};\theta_{2},n_{2}),\ldots,h_{{\color[rgb]{0,0,0}{i,N}}}(p_{N};\theta_{N},n_{N}))-\gamma_{i}r_{i},\\ p_{i}^{\prime}&=\varepsilon\left(\kappa_{i}r_{i}-\delta_{i}p_{i}\right),\end{split} (1) where (\cdot)^{\prime}=\frac{d}{dt} , i=1,\ldots,N , r_{i}\geq 0 denotes the concentration of transcribed mRNA associated with a particular gene i , p_{i}\geq 0 denotes the concentration of the corresponding translated protein, \gamma_{i},\varepsilon,\kappa_{i},\delta_{i},\theta_{i},n_{i} are strictly positive parameters, each h_{{\color[rgb]{0,0,0}{i,j}}} is a Hill function of the form h_{{\color[rgb]{0,0,0}{i,j}}}=h^{+} or h_{{\color[rgb]{0,0,0}{i,j}}}=h^{-} , where \displaystyle h^{-}({\color[rgb]{0,0,0}{p}};{\color[rgb]{0,0,0}{\theta}},{\color[rgb]{0,0,0}{n}})=1-h^{+}({\color[rgb]{0,0,0}{p}};{\color[rgb]{0,0,0}{\theta}},{\color[rgb]{0,0,0}{n}})=\frac{{\color[rgb]{0,0,0}{\theta}}^{{\color[rgb]{0,0,0}{n}}}}{{\color[rgb]{0,0,0}{p}}^{{\color[rgb]{0,0,0}{n}}}+{\color[rgb]{0,0,0}{\theta}}^{{\color[rgb]{0,0,0}{n}}}}, (2) and each function F_{{i}} is a polynomial in its arguments. We refer to e.g. [ 10 , 46 ] for details, and to Figure 1 for a graphical representation of the Hill functions h^{\pm} . Figure 1: The Hill functions h^{-}(p_{i};\theta_{i},n_{i}) and h^{+}(p_{i};\theta_{i},n_{i}) are shown in in the left and right panels respectively, for \theta_{i}=1 and three different values of n_{i} . Both functions converge do a discontinuous step function when n_{i}\to\infty , as in ( 3 ). Figure 2: Conceptual schema for the combined approach to model reduction in the (generally high dimensional and highly nonlinear) GRN model ( 1 ). Applying partial reductions in different orders (or equivalently, taking a different path from the original model ( 1 ) to the final reduced model) can lead to different reduced models, and therefore to different predictions. Direct analysis or simulation of the full network dynamics using system ( 1 ) is difficult because it is both (i) high dimensional and (ii) highly nonlinear. In order to deal with these obstacles in practice, mathematical modellers study simpler reduced models which are lower dimensional, less nonlinear, or both. The conceptual schema is illustrated in Figure 2 . Notice that the high dimensionality and high nonlinearity of the original model ( 1 ) are dealt with via separate reductions, referred to as partial reductions in the figure and its caption. Depending on the context, it may be sufficient to work with a partially reduced model which is ‘only’ lower-dimensional or less nonlinear, but in many cases the overall complexity of the problem requires a lower-dimensional, less nonlinear reduced model is only obtained after applying both of these reductions together; we refer to [ 46 ] for a review and investigation into the validity of the reductions that are often applied in practice. This paper will focus on the utility and validity of combined reductions of this kind. In mathematical terms, the partial and independent dimensional/nonlinearity