The CellModeller Project

CellModeller is a python module for 2D multicellular analysis and computing. It consist fundamentally of a generic data structure (multiscale graph representation) and elementary methods or operations to both query and modify such objects (e.g get the neighbour of a cell or modify the concentration of its constituents). The behaviour and properties of each entity in the structure (vertex, wall, cell ...) can be controlled through user specified models, using python scripting language.

Modelling the Choleochaete morphogenetic system

The Coleochaetales form a small group of microscopic algal species that are found in freshwater. C. orbicularis and C. scutata grows as discoid multicellular colonies with a simple meristem structure. The colonies adhere to a substrate and undisturbed colonies can maintain a circular shape up to several millimetres in diameter, as a result of precisely coordinated sequences of anticlinal and periclinal divisions.

Figure1: Different morphologies are observed during the development of choleochaete. These patterns can be reproduced by a model varying the values of cell wall bending properties.

Cell modeller has been used to analyse the morphogenetic processes driving the development of colonies. The simplicity of Coleochaete structures and developmental patterns made an ideal system for coupling both physical and biological aspect of multicellular development: the segmentation of live imaging data was used to derive simple cell division rules, data on the expansion of colonies could be used to derive a simple biomechanical model of cell expansion, and CellModeller could be used to investigate biological and physical factors influencing the development of colonies.

 animation showing the patterns of cell division and expansion during the development of Coleochaete resulting from the simulation of the model

input files for data and models from the Coleochaete studies


Modelling of the trichome patterning system

During leaf development, the differentiation of epidermal cells into trichomes (leaf hairs) occurs in precise patterns and frequencies. Recent work on the Arabidopsis trichome system has identified the role of various genes involved in the regulation of trichome patterning. GL1 (GLABRA1) and TTG1 (TRANSPARENT TESTA GLABRA1) are two transcription factors required to the formation of trichomes. GL3 is a positive regulator of trichome fate which associates with GL1 and TTG1 to form a complex that activates genes associated with trichome fate which we reduce here to the single GL2 gene. The lateral movement of inhibitors regulates the activity of this complex. In non trichome cells, the inhibiting proteins TRY (TRIPTYCHON) and CPC (CAPRICE) compete with GL1 and an inactive complex is formed (TRY or CPC/GL3/TTG1).


Figure 2: modification of the modelled genetic regulatury network of the epidermis cells modifies the phenotypes of trichome distribution on the leaf.

CellModeller have been used to study a simplified plausible regulatory network representing interactions between these regulatory genes. The behaviour of this simple model showed that not only realistic patterns of trichome positioning could be obtained, but also that it is possible mutate the model and analyse the phenotypes of virtual mutants. The cpc mutant showed an increase in trichome density as observed experimentally. The try virtual mutants had clusters of trichomes. Real try phenotypes generate also clusters of trichomes , although generally made of two cells in real mutants. GL3 overexpression increased trichome density as observed experimentally.

Animation of the dynamics of epidermis cell fate by simulating the trichome genetic model in a growing plate of cells


Input file for trichome patterning model

Mechanical interaction within tissues

An example of cell/cell mechanical interactions has been performed in CellModeller by simulating simple cases of outgrowth. In the shoot apical meristem for example, initiations of primordium is accompanied by cell proliferation under the L1 layer initiate the primordia. Expanding cells remain largely adherent to surrounding tissues, and the mechanical behaviour of all tissues influences the kinematics of expansion in the emerging meristem. According to (Selker et al., 1992) a possible way for the meristem to develop a localized bulge is that a localized pressure and/or a wall softening in epidermal cells initiate the primordia. This was modelled by a simple architecture of cells consisted of three different tissues: the first tissue was constituted of one layer of cells and represented the L1 layer. The second tissue was composed of 3 layers of cells under the L1 layer (orange). At the beginning of the simulation, 4 cells of the second tissue enter a proliferation stage (third tissue in green). The wall properties (viscosities and moment of inertia) are set at the level of the tissue:

Figure 3: Different types of cells organized in tissues can generate complex patterns of stress and deformation in tissues. Here, a group of internal cells are characterized by higher expansion rate. They induce the formation of an outgrowth.


Input file for outgrowth model.


In order to facilitate the development of models, analyses and visualisation techniques, CellModeller provides a large range of additional tools and methods:

- Rendering windows and various output methods to visualize the dynamical cellular architectures.

- Drawing and plotting were integrated using the Matplotlib library, 2D plotting library which produces publication quality figures.

- A python scripting environment to develop phenomenological models of the dynamics, for any entity in the multicellular structure.

- A Python shell (PyCrust) to debug the models and browse interactively the result of computations at any time of the simulation.

- Numerical methods using scientific package such as SciPy or Numeric, which are "Matlab like" array/matrix libraries.


Download the help file



Download source files