Publication
Microorganisms are a key component in the chain of life. They are essential for agriculture, produce a large proportion of oxygen, and play a central role in the cycle of elements. Microorganisms are widely used in the production of food and alcoholic beverages, pharmaceuticals, industrial chemicals, insecticides, and waste treatment. In nature, microorganisms grow in communities called microbiomes. Microbiomes colonize human and animal skin and intestinal tracts, plant leaves and roots, soil, and water. Most importantly, the composition of the microbiome is closely linked to the health of the host. Thus, there is a need to understand the underlying principles of microbiome formation, function, and evolution.
Microbial communities depend heavily on the metabolic capabilities of individual community members, as well as interactions among individuals and between the microbiome and the host or habitat, to form, evolve, and adapt. The inherent complexity of metabolic networks poses a significant challenge to the study of metabolism, and the challenge becomes even greater when more organisms and emerging interactions are considered. Therefore, the use of metabolic models and computational methods is essential for efficient experimental design and analysis of large data sets. Technological advances in sequencing and omics data processing have made it possible to generate genome-scale metabolic models for cells. These models are valuable for elucidating the genotype-phenotype relationship of cells and for generating new hypotheses to guide experimental design and have found application in basic biology, medicine, biotechnology, and more recently in microbiome research.
In this thesis, we present models, workflows, and approaches to help address existing challenges in microbiome research. We developed the NICEgame workflow to efficiently characterize metabolic gaps in genome-scale metabolic models. We harnessed the potential of this workflow and published methods to build curated genome-scale model collections for the phyllosphere of A. thaliana and a synthetic soil community. We used these model collections to predict the outcomes of metabolic interactions for pairs of strains and small synthetic communities for the phyllosphere and the metabolic requirements of the members of the synthetic soil community. Furthermore, we generated a genome-scale model for Salmonella Typhimurium specific to the metabolism of the pathogen in the mouse gut. We used this model to explore the essential metabolic genome of the bacterium and to relate gene essentiality to substrate availability. Finally, we presented the NIS workflow for the systematic comparison of the metabolism of different organisms.
The work in this thesis demonstrates the potential of metabolic models and computational methods to help navigate the inherent complexity and uncertainty of microbial communities, making a step forward in gaining a deeper understanding of microbiomes and using this knowled