Nous développons une approche générique pour comprendre comment différents régimes alimentaires peuvent influencer la qualité et la composition du lait. Cette question s'intègre dans le cadre du Flux Balance Analysis (FBA), qui consiste à analyser un réseau métabolique en optimisant un système de contraintes linéaires. Nous avons proposé une extension du FBA pour analyser la transformation des nutriments en intégrant des hypothèses biologiques utilisées par différents modèles numériques dans un modèle générique de la glande mammaire. Notre méthode permet de quantifier les précurseurs qui interviennent dans la composition des sorties du système, en calculant des contributions des entrées dans les sorties [AIO]. A l'aide de cette approche, nous avons montré que la transformation des nutriments du lait ne peut pas être modélisée par l'optimisation d'une combinaison linéaire des flux des réactions sur un modèle du métabolisme mammaire. Pour étudier plus précisément la flexibilité d'un réseau métabolique, nous avons proposé un algorithme efficace de recherche locale pour calculer les valeurs extrémales des coefficients des AIOs. Cette approche permet de discriminer les traitements sans formuler d'hypothèses sur le comportement interne du système. / This thesis proposes a generic approach to understanding how different diets affect the quality and composition of milk. This question is addressed in the framework of Flux Balance Analysis (FBA), which considers metabolic network analysis as an optimization issue on a system of linear constraints. In this work, we extended FBA to take into account nutrients transformation by incorporating general assumptions made by various numerical methods in a generic stoichiometric model of the mammary gland. Our method tries to quantify the precursor composition of each system output and to discuss the biological relevance of a set of flux in a given metabolic network. The composition is called contribution of inputs over outputs [AIO]. Using this method on the mammary metabolism, we could show that nutrients transformation cannot be properly modelled by optimizing a linear combination of reactions fluxes in the mammary gland model. In order to further investigate metabolic network flexibility, we have proposed an efficient local search algorithm computing the extremal values of AIO coefficients. This approach enables to discriminate diets without making any assumption on the internal behaviour of the system.

The increasing spectrum of multidrug-resistant bacteria has caused a major global public health concern, necessitating the discovery of novel antimicrobial agents. Additionally, recent advancements in the use of microbial cells for the scalable production of industrial enzymes has encouraged the screening of new environments for efficient microbial cell factories. The unique ecological niche of the Red Sea points to the promising metabolic and biosynthetic potential of its microbial system. Here, ten sequenced Bacilli strains, that are isolated from microbial mat and mangrove mud samples from the Red Sea, were evaluated for their use as platforms for protein production and biosynthesis of bioactive compounds. Two of the species (B.paralicheniformis Bac48 and B. litoralis Bac94) were found to secrete twice as much protein as Bacillus subtilis 168, and B. litoralis Bac94 had complete Tat and Sec protein secretion systems. Additionally, four Red Sea Species (B. paralicheniformis Bac48, Virgibacillus sp. Bac330, B. vallismortis Bac111, B. amyloliquefaciens Bac57) showed capabilities for genetic transformation and possessed competence genes. More specifically, the distinctive biosynthetic potential evident in the genomes of B. paralicheniformis Bac48 and B. paralicheniformis Bac84 was assessed and compared to nine available complete genomes of B. licheniformis and three genomes of B. paralicheniformis. A uniquely-structured trans-acyltransferase (trans-AT) polyketide synthase/nonribosomal peptide synthetase (PKS/NRPS) cluster in strains of this species was identified in the genome of B. paralicheniformis 48. In total, the two B. paralicheniformis Red Sea strains were found to be more enriched in modular clusters compared to B. licheniformis strains and B. paralicheniformis strains from other environments. These findings provided more insights into the potential of B. paralicheniformis 48 as a microbial cell factory and encouraged further focus on the strain’s metabolism at the system level. Accordingly, a draft metabolic model for B. paralicheniformis Bac48 (iPARA1056) was reconstructed, refined, and validated using growth rate and growth phenotypes under different substrates, generated using high-throughput Phenotype Microarray technology. The presented studies indicate that several of the isolated strains represent promising chassis for the development of cell factories for enzyme production and also point to the richness of their genomes with specific modules of secondary metabolism that have likely evolved in Red Sea Bacilli due to environmental adaptation.