< Back to previous page

Project

Biology driven Complexity Reduction of Individual based Models

Microbial communities during artisanal cheese production and industrial starter cultures used in large-scale cheese manufacturing largely determine the properties of the cheeses produced, together with the ripening conditions. The aim of Jian Wang's thesis is to generate predictive in silico models based on in vitro experimental data obtained during cheese production. High throughput experimental cheese analogues as developed for semi-hard cheese (NIZO: Bachmann et al. 2009) are ideally suited to study multiple parameters during fermentation and ripening to obtain relevant data that allow for building predictive models. Together with genome/transcriptome/metabolome-based data sets of starter cultures, prediction of key traits of strains in relation to outcomes of cheese properties are foreseen, allowing for rational combinatorial selection. The high throughput models will be used for artisanal and new types of cheeses, namely those based on plant proteins. Throughout this work package the power and limitations of two predictive microbial modelling paradigms will be studied: on the one hand Individual-based Models (IbM) and on the other hand Partial Differential Equations (PDEs). In the context of multi-organism populations, Individual-based Model (IbM) approaches can cope with the spatial distribution that will be useful to describe cell-to-cell interaction. A typical challenge is to provide a bridge between the (computationally very intensive) IbMs and population models which typically take the form of PDEs. Currently, PDE system descriptions, called the “Eulerian model”, are limited to the overall population evolution. On the other hand, Stochastic Differential Equations, called the “Lagrangian model”, unfortunately, present an increasing complexity with the number of individuals. Therefore, the following overall modelling challenges are addressed for cheese ecosystems modelling by Jian Wang (KUL): (i) development of IbMs considering complex mechanical and dynamical properties of cheese ecosystems and integrating metabolic network information at individual cell level as generated by WP1 of E-MUSE (initial work along these lines has been reported by KUL: Tack et al., 2017), (ii) lowering the computational burden inherent to IbMs, by developing PDE models starting from these IbMs, while maintaining model accuracy at a prespecified level, and (iii) parameter identifiability study of the generated mathematical models in view of the available experimental data, including model parametric sensitivity analysis in order to detect possible over-parameterization.
Date:10 Aug 2022 →  Today
Keywords:Individual-based model, Population-based model, Predictive modelling, Microbial communities behaviour, Cheese ecosystems, System biology
Disciplines:Computational evolutionary biology, comparative genomics and population genomics, Microbiology not elsewhere classified, Biochemistry and metabolism not elsewhere classified, Reaction kinetics and dynamics, Statistical mechanics, structure of matter, Thermodynamics, Fluid physics and dynamics, Numerical analysis, Game theory, economics, social and behavioural sciences
Project type:PhD project