Mind and Body: Statistical Models for Two Biological Networks
This talk exemplifies how statistical models for networks can learn associations between species in biological systems. In the first application, we model the default mode network in the human brain as a Copula Generated Random Graph (CGRG). A term measuring reciprocity in the regulatory network reveals hierarchy in the connectome breaks down as neurodegenerative disease progresses. As a result, the model can be used to detect Alzheimer’s disease. In the second application, we estimate a latent network structure in a protein signaling network. The model incorporates theoretical knowledge about systems biology. A sequential Monte Carlo algorithm provides an efficient way to sample from the posterior distribution of network structures.