Joint inference and alignment of genome structures enables characterization of compartment-independent reorganization across cell types

Abstract Cell-type-specific chromosome conformation is correlated with differential gene regulation. Broad compartmentalization into two compartments (A & B) is proposed to be the main driver of cell-specific chromosome organization. However it is unclear what fraction of chromosome conformation changes between cell types and conditions is independent of changes in compartmentalization and whether any such compartment-independent reorganization is functionally important. We developed MultiMDS to jointly infer and align 3D chromosomal structures, thereby enabling a quantitative comparison of locus-specific changes across Hi-C datasets. We compared Hi-C datasets from yeast, which lack compartmentalization, grown with and without galactose. These comparisons confirmed known relocalizations as well as identifying additional examples. We also compared mammalian datasets across a variety of cell lines. We found a consistent enrichment for changes along the A/B compartment (nuclear interior/nuclear periphery) axis, even when comparing the same cell type from different individuals. Despite the prevalence of compartment changes, we consistently find compartment-independent relocalizations of loci that are within the A compartment in both compared cell types. Some such intra-compartment relocalizations involve loci that display enhancer-associated histone marks in one cell type and polycomb-associated histone marks in the other. MultiMDS thus enables a new way to compare chromosome conformations across two Hi-C datasets. Availability https://github.com/seqcode/multimds