We present a novel, multi-level, computational framework for the synthesis and analysis of mammalian metabolism from the point of nutrient uptake, through a complex network of catabolic (energetic) and anabolic (biosynthetic) reactions, and to the point at which a large protein product is synthesized and possibly secreted. This framework is predicated upon dynamic systems theory, and adopts a perspective of input and output relationships as determined through a unique network state. This network state is, in turn, determined through the simultaneous optimization of a universal linear network flux-balance subject to certain unique non-linear constraints. Further, we will demonstrate the ability of this framework to identify valid, yet non-intuitive, metabolic chemistry when tasked with the optimization of cellular processes relative to the maximum yield of a protein product. We use this framework to probe for bottlenecks in synthesis of large non-native proteins (i.e. monoclonal antibodies) from mammalian cell hosts. Finally, to integrate this computational effort with the experimental laboratory, we explore specific challenges in the engineering of cell lines to adopt this non-intuitive metabolic chemistry. It is our hope that such engineered cell lines will demonstrate the improved productivity predicted by our model.