| dc.description.abstract |
In December 2019, the world was hit by the global SARS-CoV-2 pandemic. Two years later, the number of infections and deaths is still increasing, affecting everyday’s life. Although several vaccines have been successfully approved for SARS-CoV-2, therapeutic strategies are still minimal. As viruses rely on the host’s metabolism for replication, analyzing the viral reprogramming of the host cells might reveal potential antiviral targets. Such metabolic alterations can be evaluated and analyzed using genome-scale metabolic models (GEMs). These models represent large metabolic networks that connect metabolites with biochemical reactions facilitated by proteins and encoded by genes. With the help of genomic information, the so-called genotype, we can create metabolic models that can predict the phenotypic behavior of an organism. However, these GEMs can be used to analyze virus-host interactions and predict potential antiviral targets and understand the genotype-phenotype relationship of pathogens and commensals such as Staphylococcus aureus. High-quality models with a high predictive value help us to better understand an organism, determine metabolic capabilities in health and disease, identify potential targets for treatment interventions, and analyze the interplay between different cells and organisms. Such models can answer relevant and urgent questions of our time quickly and efficiently and become an indispensable constituent in future research.
In my thesis, I demonstrate (I) how the quality and predictive value of an existing genome-scale metabolic model can be assessed, (II) how high-quality genome-scale metabolic models can be curated, and (III) how high-quality genome-scale metabolic models can be used for model-driven discoveries.
All three points are addressed in the context of pathogens and commensals in the human respiratory tract. To assess the quality and predictive value of GEMs, we collected all currently available models of the pathogen Staphylococcus aureus, which colonizes the human nose. We evaluated the models concerning their validity, compliance with the FAIR data principle, quality, simulatability, and predictive value. Using high-quality models with a high predictive value enables model-driven hypotheses and discoveries. However, if no such model is available, one needs to curate a high-quality model. For this purpose, we developed a pipeline that focuses on the model curation of nasal pathogens and commensals. This pipeline is adaptable to incorporate other tools and bacteria, pathogens, or cells while maintaining certain community standards. We demonstrated the applicability of this pipeline by curating the first model of the nasal commensal Dolosigranulum pigrum. We showed how to use high-quality GEMs for model-driven discoveries by identifying novel antiviral targets. To do so, we virtually infected human alveolar macrophages in the lung with SARS-CoV-2. |
en |
