Classical and Deep Kinetic Modelling

Svetlana Volkova

Research output: Book/ReportPh.D. thesis

110 Downloads (Pure)


Understanding of the cell metabolism is the ultimate goal of biochemistry. Understanding of the system means that we can model and predict the future states of the system. One way to model metabolism is to use kinetic models describing changes in metabolite concentrations over time. We demonstrated how application of kinetic modelling can advance our understanding of cell response to drugs, metabolism in health and disease and ability to select the target for biotechnological efforts. Furthermore, we developed a hybrid approach that will lead a foundation to new methods in kinetic modelling, potentially overcoming some setbacks in current methods.

Application of many drugs is hindered by their side effects that can be severe. We’ve constructed kinetic models describing Red Blood Cells and Platelets metabolism and integrated metabolomic experimental data to find what metabolic pathways are changing during uptake of aspirin. Our results highlight complex phenomena underlying this response and potentially will bring us closer to understanding the mechanism of side effects of the aspirin.

Red Blood Cells are a very common planned or collateral target of drugs. Another drug that is commonly used and also potentially can lead to anemia is ribavirin. Chemical features of ribavirin make it a potent modulator of purine metabolism particularly in Red Blood Cells and we’ve created a detailed kinetic model linking purine pathways with ribavirin action. Our results show inconsistencies within previously published and commonly assumed mental models of ribavirin modes of action and suggest experiments that need to be done to clarify the mechanism behind ribavirin response.

Universal nature of kinetic modelling principles allows us to apply modelling to challenges in chemical bioproduction. Acetyl-coenzyme A is a precursor for a number of high-value compounds therefore it is desirable to design such strains that can produce it in bigger quantities. Due to the high degree of regulation within metabolic pathways kinetic models are a natural choice to create a formal model. We’ve created the Pseudomonas putida central carbon metabolism model to identified gene-targets to boost of acetyl- CoA production with metabolic control analysis and performed the analysis of omics data to describe the engineered cell lines.

Development and analysis of kinetic models is a highly nontrivial process and is inherently limited in a number of ways - there are requirements for knowledge of kinetic parameters that are hard to obtain, kinetic laws may not be flexible enough in some cases and simulation speed can be a potential bottleneck. We propose a different way of formulating metabolic models within Graph Deep Learning formalism. Our experiments show promising features of such formalism and lay path for the flexible and performant hybrid models making simulations and analysis of very complex systems possible in the future.
Original languageEnglish
PublisherTechnical University of Denmark
Number of pages186
Publication statusPublished - 2021


Dive into the research topics of 'Classical and Deep Kinetic Modelling'. Together they form a unique fingerprint.

Cite this