Chromatin accessibility is associated with gene regulatory elements and gene transcription. It has been found to be tissue specific and in more recent studies, cancer specific. Identification of the state/function of regulatory elements and their target genes in patients using chromatin accessibility and transcriptomic data poses a computational challenge. We developed a method called DGTAC (Differential Gene Targets of Accessible Chromatin) to solve these problems using a machine learning model trained on 3D chromatin contact data from cell lines. We applied DGTAC to a large patient chromatin accessibility dataset of 400 cancer patients across 22 cancer types and identify novel cancer-type and subtype-specific enhancer-gene connections for known cancer genes, which we validated using CRISPR. Using the regulatory networks constructed using DGTAC, we identify clusters of patients representing shared underlying cancer biology from disparate cancer types and disease-specific subtypes as well as the key regulators in these clusters. We leverage these newly identified key regulators and publicly available datasets to identify cluster-specific, novel therapeutic targets and validate inhibitors of these targets in model cell lines. Thus, this thesis presents new computational approaches that can be used to analyze the ATAC-seq and RNA-seq datasets that can now readily be obtained from patient biopsies for translational research.