The dramatic decrease in sequencing and super-computing costs has enabled the generation of very large-scale datasets and the use of advanced algorithmic solutions applied to those datasets to achieve a better understanding of complex diseases. In this thesis, I integrate multiple modalities of data and apply rigorous statistical and mathematical modeling approaches to analyze these data and create data-driven hypotheses. I start with the exploration of the reproducibility of a commonly used modeling approach, probabilistic causal Bayesian networks, and then application of this modeling method to two complex diseases: Coronary Artery Disease (CAD) and food allergy. Using integrative approaches and a large CAD cohort, I detected downstream effects of GWAS genes via cis- and trans- eQTLs and identified a liver-specific regulatory sub-network that inversely affects plasma cholesterol and blood-glucose levels. Applying a similar framework to longitudinal measurements in peanut allergy patients with and without being challenged with peanut exposure, I found specific transcriptomic changes and highlighted novel regulators of the allergy response. This work emphasizes the importance of using integrative approaches to uncover novel regulators of complex human disease.