DECODING COMPLEXITY:HARMONISATION OF MODELLING PARADIGMS

Rationale Computational modelling is a cornerstone of Systems Biology, enabling mechanistic insights into complex biological systems. Computational models are crucial for addressing critical questions about system evolution and deciphering system connections. To gain recognition within the biological and clinical communities, these models must offer a holistic view of the entire system while simultaneously allowing inspection of its components at varying levels of granularity. Results This PhD research advances hybrid modelling approaches that integrate Ordinary Differential Equations (ODEs) with Genome-ScaleMetabolicModels (GEMs), combining fine-grained dynamic and coarse-grained steady-state representations. This integration, termed harmonisation, allows bidirectional coupling between ODE-based cellular dynamics and constraint-based Flux Balance Analysis (FBA), enabling time-resolved simulations of context-specific metabolic behaviours. This thesis presents UnifiedGreatMod, an extension of the GreatMod framework, which formalises hybrid models using Extended Stochastic Petri Nets (ESPNs). It enables functional studies by combining the analysis of the stability at multiple levels of the system with its fluctuating conditions. This approach helps investigate the functional relationships and dependencies among biological entities. The proposed paradigm was implemented into an open-source general modelling framework to define user-defined analysis workflows and in which a high-level graphical formalism is exploited to simplify the model creation phase. These approaches capture key phenomena, such as metabolite exchange, growth, interspecies interaction, and community assembly, through an integrated, and mechanistically interpretable computational framework. Harmonisation strategies are demonstrated through a progression of use cases: from single-organism metabolic regulation (e.g., Clostridioides difficile infections, capturing pathogen-host-environment dynamics), to transcriptome-constrained simulations (e.g., Escherichia coli diauxic growth), and finally to more complex microbial ecosystems involving multiple interacting species (e.g., SIHUMIx and minimal synthetic consortia). Implications The strategies presented here have general implications for Systems Biology and translational research. Special emphasis is placed on developing and simulating community-scale microbiota models. Microbial communities are central to human health, agriculture and industry, and their complexity is derived from species diversity and complex metabolic interactions. The methodology includes automatic model compilation, parameter-driven dynamical model solving, and support for multiple linear programming solvers. The central outcome is an end-to-end, scalable toolchain for multi-scale microbiota modelling, supporting the data integration. The resulting platform is designed to be modular, extensible, and reproducible, supporting seamless coordination of diverse model components and data sources. Target modelling was presented to corroborate the framework’s ability to simulate multiscale processes such as intracellular metabolism, mirobial population-level dynamics, and environmental feedbacks under steady-state and time-varying conditions.