The interplay between different genomic and epigenomic alterations lead to different prognoses in cancer patients. Advances in high-throughput technologies, like gene expression profiling, next-generation sequencing, proteomics, and fluxomics, have enabled detailed molecular characterization of various tumors, yet studying this interplay is a complex computational problem.Here we set to develop computational approaches to identify and study emerging challenges in cancer metabolism and genomics. We focus on three research questions, addressed by different computational approaches: (1) What is the set of metabolic interactions in cancer metabolism? To this end we generated a computational framework that quantitatively predicts synthetic dosage lethal (SDL) interactions in human metabolism, by developing a new algorithmic-modeling approach. SDLs offer a promising way to selectively kill cancer cells by targeting the SDL partners of activated oncogenes in tumors, which are often difficult to target directly. (2) What is the landscape of metabolic regulation in breast cancer? To this end we established a new framework that utilizes different data types to perform multi-omics data integration and flux prediction, by incorporating machine learn- ing techniques with Genome Scale Metabolic Modeling (GSMM). This enabled us to study the regulation of breast cancer cell line under different growth conditions, from multiple omics data. (3) What is the power of somatic mutations derived from RNA in estimating the tumor mutational burden? Here we develop a new tool to detect somatic mutations from RNA sequencing data without a matched- normal sample. To this end we developed a machine learning pipeline that takes as input a list of single nucleotide variants and classifies them as either somatic or germline, based on read-level features as well as position-specific variant statistics and common germline databases. We showed that detecting somatic mutations directly from RNA enables the identification of expressed mutations, and therefore represent a more relevant metric in estimating the tumor mutational burden, which is significantly associated with patient survival. In sum, my work has been focused around developing computational methods to tackle different research questions in cancer metabolism and genomics, utilizing various types of omics data and a variety of computational approaches. These methods provide new solutions to some important computational challenges, and their applications help to generate promising leads for cancer research, and can be utilized in many future applications, analyzing novel and existing datasets.