Multicellular organisms are composed of many different cell types. All such cells arise from a single cell--- the zygote--- and acquire the various cell fates seen in adult organisms. The different cell types are characterized by distinct, cell-fate-specific gene expression patterns. Cells of different types can also exhibit varying metabolic states depending on their intrinsic needs and the nutrient microenvironment. Both during development and in adult organisms, cell-fate choice is tightly controlled, and its dysregulation is known to contribute to many pathologies, including cancer. In this thesis, I describe our simulations-based efforts to identify certain general principles underlying cell-fate choice. Throughout, I discuss how such regulation can go awry in a disease like cancer, leading to the emergence of aberrant cell fates. First, I describe a spin glass-based theory of minimal frustration in regulatory networks implicated in cell-fate choice, and show that the minimal frustration property is key to the robust establishment and maintenance of biological cell-fates. The minimal frustration property is also crucial to the success of various systems biology models of cell-fate choice. Next, I present two models concerning noise in cell-fate choice--- a mechanical model of DNA supercoiling-mediated transcriptional noise and a coarse-grained model of noise in partitioning during cell division that can create and maintain a phenotypically heterogeneous population. Finally, I describe a mechanistic model of the key metabolic pathways active in tumors and other fast-proliferating cells. Our model recapitulates tumor cell behavior across contexts and makes useful predictions concerning the ways tumor cells can evade metabolic therapies. Overall, this thesis describes multiple examples of how physical and systems biology-based approaches can be leveraged to understand the key principles underlying cell-fate choice across biological contexts.