Introduction for day 2
Professor Thomas Curtis, Newcastle University, UK
Dr Jane Fowler, Simon Fraser University, Canada
Individual-based modelling: what have I learnt?
Dr Jan Ulrich Kreft, University of Birmingham, UK
For engineering systems, it is important to be able to predict the behaviour of systems before they are being engineered as prediction can avoid costly mistakes and enables optimization of systems. Mathematical models enable true predictions rather than interpolations or extrapolations from measurements of the same system, if they are mechanistic. (A similar case can be made for causal inference.) In the case of complex systems, mechanistic models have to be bottom-up models that describe the characteristics of the parts and how they interact with each other, to predict how system level behaviour emerges from these interactions. A quintessential type of bottom-up models are individual-based models, where the ‘parts’ are individual organisms, in the case of microbes often but not necessarily single cells. But how far down the scale of organization should one go when modelling individual organisms? Use Monod kinetics or incorporate gene regulatory and stoichiometric metabolic models or dynamic metabolic models? All the way to physical laws from thermodynamics and mechanics? Drawing on two case studies, the question of how to cope with the huge diversity of microbial ecotypes with their huge phenotypic flexibility, interacting with others in spatially structured and temporally fluctuating environments will be discussed with an emphasis on metabolism. The case studies are based on competing, abstract strategies rather than detailed implementations of particular manifestations of these strategies, and this choice has advantages and disadvantages. One is based on a trade-off between specific growth rate and growth yield, demonstrating the benefits of higher yields in biofilms. The second is based on a trade-off between allocating resources into repair of damaged materials or into growth and reproduction, while getting rid of damage by asymmetric damage segregation causing ageing.
Integrating metabolic versatility into spatially-explicit models of soil bacterial life
Benedict Borer, ETH Zürich, Switzerland
Soil is a harsh and dynamic environment for bacterial cells due to nutrient diffusion and dispersal limitations following episodic wetting events. Nevertheless, soil hosts unparalleled diversity of bacterial life where metabolic versatility is key to their success, enabling them to exploit diverse growth strategies on a wide range of resources. Soil bacterial life is governed by localized nutrient conditions at the microscale, giving rise to complex metabolic landscapes that shape bacterially-mediated processes ranging from soil nutrient cycling to greenhouse gas emissions. Interestingly, these nutrient landscapes can trigger fundamentally different growth strategies even for the same species when residing in close proximity. Most mathematical models are currently unable to capture such versatility and local adaptation, calling for a more nuanced representation of bacterial metabolism and the soil physical structure. Benedict Borer reports a mathematical framework that considers individual bacterial cell dispersal and interaction with nutrient diffusion fields in a spatial context that embrace metabolic versatility using flux balance analysis based on genome scale metabolic networks. Benedict investigates the spatial organisation of a synthetic bacterial community in artificial pore networks and reveal mechanisms promoting spatial segregation that enable coexistence. Representing the aqueous phase architecture of soil at the cell scale offers unprecedented opportunities to interrogate bacterial life in complex habitats that is typically veiled by soil opacity.
Upscaling and statistical emulation of individual based models
Professor Darren Wilkinson, Newcastle University, UK
Modelling entire ecological communities of bacteria in large, complex open environments is extremely challenging. The NUFEB project involves a multi-disciplinary team of researchers developing methods and software for multiscale modelling of open engineered biological systems at scale. The group's exemplar project is focused on the modelling of wastewater treatment systems which have macro-scale characteristics arising from the micro-scale features of up to 10^18 individual interacting bacteria. They have developed an individual based model of bacterial communities in an active fluid environment, and they can use this to understand the small-scale features of the system, considering volumes containing up to a few million bacteria. At the system scale the group are developing continuum models which capture essential macro-scale properties. They propose a novel technique for coupling the two models to produce a multi-scale model by embedding fast statistical emulators of the individual based model into the macro-scale model. This talk will outline the current state of this work in progress on this ongoing project.
Computational modelling of microbial communities
Professor Ines Thiele, National University of Ireland, Ireland