Multi-stage volume exclusion models for cell proliferation

John Carlo Dimaculangan, Cameron A. Smith, Christian A. Yates · 2026

Read Multi-stage volume exclusion models for cell proliferation by John Carlo Dimaculangan, Cameron A. Smith, Christian A. Yates (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-2604-19852.txt.

Introduction

Dimaculangan, Smith, and Yates model how cells multiply and move when crowded, using stochastic on-lattice simulations in which each cell cycle runs through several stages. A novel myopic rule lets cells sense their local surroundings before dividing. For each simulation they derive a matching continuum equation, then compare the two descriptions across growth-to-confluence assays and invasion waves, showing how division rules shape population spread.

Catalog record

CREATORJohn Carlo Dimaculangan, Cameron A. Smith, Christian A. Yates
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE89691
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2604.19852

Browse the full library · About the Tower of Babel

Full text

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

Multi-stage volume exclusion models for cell proliferation John Carlo Dimaculangan, Cameron A. Smith, Christian A. Yates Abstract Cell proliferation and cell movement are fundamentally stochastic processes which lead to variability in the growth and spatial structure of cell populations in many biological settings, such as cell invasion, wound healing, and tumour growth. We develop stochastic, on-lattice agent-based models (ABMs) which incorporate volume exclusion, random movement, and multi-stage representations of the cell cycle. The multi-stage framework enables a more realistic representation of true cell cycle time distributions. We also introduce a novel form of myopic behaviour, in which cells sense their local environment when attempting to proliferate. For each ABM, we derive a corresponding continuum partial differential equation (PDE) description under the mean-field approximation. Using numerical simulations, we investigate how different proliferation mechanisms influence population-level dynamics in both the discrete and continuum models. In particular, we consider biologically relevant contexts of growth-to-confluence assays (using uniform initial conditions) and travelling wave behaviour associated with cell invasion. We examine how the PDE solutions compare with the behaviour of the corresponding ABMs averaged over many realisations. I Introduction Cell proliferation is a fundamental biological process which underpins tissue development, immune responses 1 , wound healing 18 , and cancer growth 17 . Understanding the cell cycle, the framework which regulates when and how cell proliferation occurs, is crucial to developing biologically realistic models of such processes. The cell cycle consists of four distinct phases: \text{G}_{1} phase (Gap 1) where cells grow and prepare to replicate their DNA, S phase (Synthesis) where cells replicate their DNA, \text{G}_{2} phase (Gap 2) where cells prepare for mitosis and continue to grow, and M phase (Mitosis) where a cell divides 36 . Cells are able to temporarily arrest at key checkpoints prior to progressing into the S or \text{G}_{2} phases if they are sufficiently damaged or conditions for proliferation are unfavourable ( 34 ; 21 ) . In particular, it has been shown that crowded environments can cause a cell to become quiescent ( 28 ; 25 ) . Therefore, accurate mathematical models of such cell proliferation should incorporate density dependent behaviour within the cell cycle. In this paper, we consider two complementary approaches to modelling the cell cycle. The first approach uses on-lattice ABMs which are able to capture individual cell behaviour and stochasticity, which is inherent to cell cycle and to cell movement. ABMs can be used to understand the spatial heterogeneity which can arise within a growing population of cells. Furthermore, it is relatively simple and intuitive to implement specific cellular behaviour or cell-cell interactions within these frameworks 15 . However, ABMs have the disadvantage of a high computational cost, and limited analytical tractability 2 . The ability to accurately capture cell-cycle time distributions is essential for constructing biologically realistic models of cell proliferation. Previous work often makes the simplifying assumption that cell cycle times follow an exponential distribution ( 4 ; 29 ; 43 ) . This assumption is convenient as it allows for simulations to be carried out using Gillespie’s algorithm 14 . However, it can be shown that cell cycle times are not well represented by the exponential distribution 16 . In contrast, the multi-stage model for the cell cycle partitions the cell cycle into multiple stages (distinct from the biological phases above) with exponential waiting times, which cells must progress through prior to dividing. This allows the multi-stage model to yield cell cycle time distributions which accurately reflect experimentally observed cell cycle times 49 . Importantly, because these stages are modelled independently of each other, the multi-stage model can be implemented using Gillespie’s algorithm. ABMs incorporating both cell proliferation and cell movement have also been widely studied. Some models assume that cell cycle times follow an exponential distribution ( 5 ; 40 ; 41 ) , whereas other models adopt the multi-stage representation of the cell cycle ( 39 ; 11 ) . Under standard multi-stage models, cells choose the site of placement of the daughter cell at random from its neighbours, with the event aborted if the site is occupied. In contrast, myopic behaviour grants the ability for cells to sense their local environment. Previously, myopic behaviour has been studied in the context of cell movement by Landman and Fernando 24 . In this paper, we will continue to investigate spatially extended multi-stage ABMs and compare their behaviour to spatial ABMs which use the simpler assumption of exponentially distributed cell cycle times. We will also adapt the approach taken by Landman and Fernando and study myopic behaviour in the context of cell proliferation. We will determine how incorporating myopic behaviour into ABMs changes their behaviour, and study the effects of myopic behaviour together with the multi-stage model. The second approach we take to modelling density dependent cell proliferation and cell movement is to use continuum (non-spatial) ordinary and (spatial) partial differential equations (ODEs/PDEs) which can capture the macroscale dynamics of a system. There is well established theory (such as steady-state and bifurcation analysis) for many continuum models, and it is computationally cheaper to numerically solve a system of ODEs/PDEs in comparison to simulating an equivalent population of cells using ABMs. However, continuum models are unable to capture any stochastic behaviour within a system, and can neglect the local structure 5 . Non-spatial models of cell proliferation are typically formulated using coupled systems of ODEs ( 46 ; 13 ) . These frameworks incorporate realistic cell-cycle time distributions by adopting an underlying multi-stage structure, in which each ODE represents the density of cells in each different stage of the cell cycle. Delay differential equations (DDEs) have also been proposed ( 47 ) , in which a distributed delay kernel is used to impose specific cell-cycle time distributions. DDEs are also able to capture realistic cell cycle time distributions, but at the cost of a more complicated model. To account for spatial effects such as migration and crowding, these approaches must be extended. Classical spatial models often take the form of reaction–diffusion equations, such as FKPP-type equations with logistic growth ( 30 ) . However, in their simplest form, these models implicitly assume exponential waiting times between cell division attempts. More accurate representations of cell-cycle dynamics can be obtained by extending these formulations to systems of coupled PDEs, where each PDE represents cells in a different cell cycle stage ( 45 ; 10 ) . While such models are capable of capturing crowding effects and structured cell-cycle progression, they are typically limited to a relatively small number of stages, as the analysis and parameter inference for these systems becomes more challenging with increasing model complexity. In this paper, we derive the continuum PDE limits of our discrete ABMs under various assumptions. When these assumptions are valid, we demonstrate that the resulting continuum PDEs provide an accurate description of the average behaviour across many realisations of the corresponding ABM. We also investigate parameter regimes in which the mean-field assumption breaks down, and show that, in such cases, the growth dynamics predicted by the ABMs and their associated continuum PDE limits can differ significantly. This enables us to distinguish a priori between scenarios in which PDEs successfully capture the spatial structure of the underlying

Read the complete text