Abstract. Cystic fibrosis (CF) is a fatal genetic disease characterized by overproduction of mucus in the lungs followed by chronic lung infections. Conventional wisdom has been that most CF lung infections involve a single dominant organism, most commonly the pathogenic bacterium Pseudomonas aeruginosa. Advances in culture-independent techniques have revealed that CF lung infections are rarely mono-microbial and instead usually involve complex microbial communities, yet the interspecies interactions that drive these communities are poorly understood. Furthermore, numerous studies have demonstrated that polymicrobial infections are more difficult than mono-microbial infections to eradicate with antibiotics, leading to the concept of recalcitrant communities. The mechanisms underlying recalcitrance are thought to involve synergistic interactions between community members, but very little data are available to understand this phenomenon. Combined with the realization that many CF patients respond poorly to available antibiotic regimens compels a more detailed understanding of interspecies interactions and their impacts on antibiotic recalcitrance to improve the treatment of CF infections, as well as other polymicrobial diseases. Here, we combine big-data bioinformatics, in silico computational modeling and in vitro culture experiments to gain insights into the metabolic interactions that drive CF disease outcomes and antibiotic recalcitrance. The research will leverage an available data set of hundreds of CF patient samples that provide both bacterial composition data and clinical metadata, including measures of lung function. These samples will be clustered according to their measured compositions and metabolic capabilities predicted through computational metabolic modeling to test the hypothesis that the vast complexity of these many bacterial communities can be collapsed into a small number of model communities that capture most of the observed metabolic variability. These computational predictions will be tested by developing in vitro cell culture models that recapitulate the most important metabolic features of the in vivo polymicrobial communities (Aim 1). By applying bioinformatics and modeling to the same clinical data, we will test the hypothesis that community metabolic features drive disease outcomes and the virulence potential of these communities (Aim 2). Finally, we will interrogate the clinical data and in vitro communities to test the hypothesis that community metabolic features drive antibiotic recalcitrance and differentiate community responsiveness to antibiotics according to these metabolic features (Aim 3). Our research will yield novel insights into how complex polymicrobial communities are compositionally structured, interact metabolically, contribute to disease and respond to antibiotics. Moreover, the research will validate in vitro models that offer the potential for development of novel antimicrobial strategies to better ...