Breaking down genomic compartments into their building blocks

While our DNA sequence stores genetic information, the physical spatial organization of DNA is associated with how this information is regulated. In the past decade, an experimental method called Hi-C has revealed several genomic structures over different scales. Hi-C produces a matrix of interaction frequencies between every two points in the genome. This matrix contains multiple interaction patterns, representing different structures. A major goal is understanding how these patterns are specified molecularly, what physical structures they represent and their association with biological function. Here we present an explicit computational model for a long-range interaction pattern called genomic compartments, which is commonly attributed to a partitioning of the genome into active and inactive chromatin. We show that this pattern can be explained by a model in which each locus belongs to one out of a small number of chromatin types that interact with each other. However, unlike the common partitioning into two types, our model suggests that more types are involved in this organization pattern. We also show that chromatin types interact differently inside vs between chromosomes and observe a possible mixing of these types in the cell population. Finally, we use our modeling framework to test hypotheses regarding the molecular specification of chromatin types.