Cancer is a leading cause of death in the world, and the mechanisms that underlie this disease are still not completely understood. As cancer develops and progresses, cells undergo a diversity of mutations that sustain their rapid proliferation and the evasion of the immune system. Cancer cells alter their configuration and organization, exhibiting abnormal phenotypes and changes in functionality. The complexity of cancer lies in their heterogeneity and variability among patients, which challenges the current therapies and drug targets. In the last decades, ten hallmarks of cancer cells have been recognized, including alterations in metabolism and the signaling pathways. The sequencing of the human genome and the advances in omics data processing allowed to generate metabolic and signaling networks for human cells at a genome-scale, enlightening the detailed biochemistry and signal transduction processes occurring in human cells, and enabling to study human metabolism and signaling pathways at a systems level. However, the complexity of these networks hinders a consistent and concise physiological representation. In the field of systems biology, mathematical models and computational methods are derived to describe cellular processes based on experimental data and the biological networks. Furthermore, these models have proven to be valuable in understanding the genotype-phenotype relationship of cells and to formulate new hypotheses to guide experimental design. In this thesis, we present modeling approaches and computational methods to investigate the metabolic and signaling alterations in cancer cells and overcome the challenges arising from biological networks of such size and complexity. Firstly, we curated the thermodynamic properties for all the compounds and reactions in the human metabolic genome-scale models (GEMs) Recon 2 and Recon 3D to guarantee the consistency of the predictions with the bioenergetics of the cell. Moreover, we developed a workflow (redHUMAN) for reconstructing reduced-size models that focus on parts of metabolism relevant to a specific physiology, and we introduce a novel method to account for the cellular interactions with the extracellular medium. Using redHUMAN, we reduced the human GEMs around pathways that are altered in cancer physiology. Secondly, we applied a set of computational methods to integrate omics data into the reduced version of Recon 3D to build metabolic models for breast, colon, and ovarian cancers. These models were used to study how different cancer cells use the metabolic pathways to function and survive and how the underlying genetic deregulations affect the metabolic tasks. Thirdly, we developed a method (CONSIGN) to contextualize signaling networks to a specific type of cell under particular conditions, maximizing the consistency with experimental data. We used this method to generate a breast cancer-specific signaling network for the transcription factor MYC. Finally, we created an integrated model of signaling and metabolic models by accounting for the regulation of metabolic genes by transcription factors. We analyzed the interactions of the MYC signaling network in the breast cancer metabolic model. The work in this thesis demonstrates the potential of metabolic and signaling models to identify and infer the genetic origins and the microenvironment effects in the transformed phenotype of cancer cells, marking a step forward towards the study of drug targets and biomarkers.