Unraveling strain-specific metabolic diversity using genome-scale models: development of computational methods and applications

This PhD thesis is about the study, development, and application of computational methods for the reconstruction and analysis of genome-scale metabolic models. Specifically, it focuses on state-of-the-art methods for the simultaneous modeling of multiple bacterial strains of the same species or phylogenetically close species. Each model encapsulates the metabolic potential of a strain and can be used to simulate cellular growth under various conditions, enabling genome-based phenotype predictions. The development of a novel multi-strain reconstruction method is here reported, inspired by existing approaches and designed to overcome some of their limitations. This method has been implemented into Gempipe, a freely accessible bioinformatics tool (github.com/lazzarigioele/gempipe) which also deals with model correction and analysis, and represents the main outcome of the thesis. It has been compared against the leading alternatives, proving to be a step forward in the field of multi-strain reconstructions. Furthermore, the tool has been applied in three case studies described in this thesis. In the first one, models were used to predict a suitable growth medium for sustaining the growth of Candidatus Erwinia dacicola, an endosymbiotic bacterium known to be essential for the reproduction of the olive fly, insect that damages orchards and causes economic losses in the agrifood sector. The predicted growth medium enabled a preliminary culture of the endosymbiont, but not its isolation, nevertheless revealing previously unreported growth factors. In the second case study, models were employed to investigate intraspecific metabolic biodiversity of Lactiplantibacillus plantarum, a lactic acid bacterium commonly used in food productions, assessing an eventual correlation between metabolic potential and the ecological niche from which the strains were sampled. The implemented approach enabled the simultaneous comparison of more than a thousand strains, which were divided into metabolically homogeneous groups seemingly unrelated to the niche, consistently with a lifestyle described as nomadic. In the third case study Thauera sp. Sel9 was modeled, a bacterial strain isolated from a wastewater treatment plant and capable of producing plastic biopolymers from short-chain fatty acids. The information contained in phylogenetically related genomes, combined with a high-throughput phenotypic screening, enabled the correction of the model’s topology, a necessary starting point for prospective use aimed at increasing bioplastic yields.