In the context of global warming and eutrophication of waters, harmful algal blooms are an international concern. Warm weather and nutrients abundance feed cyanobacterial species, whose growth goes alongside biodiversity degradation, health risk for animals and humans and high economical repercussions (e.g. water treatment, recreational areas closure, impacts on agri-food). Microcystis, a genus of freshwater cyanobacteria, is described as one of the most pervasive bloom forming species. A subset of Microcystis strains' ability to produce the microcystin hepatotoxin, microcystin-LR (MC-LR) being the most harmful and common, make their bloom a certain threat. The biochemical and ecological mechanisms behind those remain relatively unknown, motivating the use of in silico systems biology approaches to propose mechanistic hypotheses. Biochemical interactions between Microcystis and heterotrophic bacteria within the phycosphere may contribute to bloom formation particularly through optimal management of nutrient resources (e.g. recycling of organic matter, cross-feeding). We aim to analyse the mechanisms underlying the metabolic activity of cyanobacteria and their associated symbionts using metabolic models. Genome-scale metabolic networks (GSMNs) gather all biochemical reactions linked with the genes of a genome, and can be reconstructed automatically and feed metabolic models to predict the behaviour of organisms, or communities in their environment. Our work’s objective decomposes into three parts i) assessing the ability of GSMNs reconstruction to capture Microcystis strains production of toxins, ii) compare the metabolic functions carried by a set of Microcystis strains iii) analyse jointly the metabolism of the cyanobacteria and their communities across several lake samples in the context of bloom formation.< Réduire