Numerous experimental assays, such as Hi-C, ChIA-PET, or PCHiC, allow us to investigate the spatial organization of the genome within the nucleus. The data generated by these experiments can be naturally represented as a graph - or a "chromatin network" - where nodes represent genomic regions and edges represent spatial proximity between them. In this study, we propose a probabilistic generative model for random graphs, specifically based on the Stochastic Block Models (SBM) class, to describe the compartmentalization of interactions observed in Hi-C data. Moreover, in our model, the connectivity properties of the nodes are modulated by biochemical quantities, such as histone modifications and transcription factor binding. We also develop a variational algorithm to perform posterior inference on the model.