Inferring networks from big biomedical data is important for understanding complex biological systems and visually exploring big data. In this talk, we highlight new inference methods for graphical models inspired by neuroimaging data and integrative genomics. First, functional brain networks are important for understanding brain organization and dysfunction in neurological diseases. In population studies, scientists seek to estimate and then test for differences in brain networks related to disease or other clinical features. We present a new framework for testing populations of graphical models and apply these methods to several functional MRI studies. Second, new genomic technologies allow scientists to profile nearly every molecular aspect of a sample; but, this data is big and of mixed types (e.g. continuous, binary, counts, etc.). We introduce new integrative graphical models that give the first general multivariate distribution for mixed data and apply these to build integrative cancer genomic networks.