Preprint: DiffDomain-Spectrum identifies reorganised topologically associating domains from sparse single-cell Hi-C data

A preprint on bioRxiv describes DiffDomain-Spectrum, a spectral statistical framework for identifying structurally reorganised topologically associating domains (TADs) between conditions or cell types using aggregated single-cell Hi-C (scHi-C) contact maps.

TADs are megabase-scale chromatin domains that constrain enhancer–promoter interactions and whose boundaries are maintained by architectural proteins such as CTCF and cohesin. Disruption of TAD structure has been linked to dysregulation of gene expression in cancer and developmental disorders. Single-cell Hi-C technologies allow chromatin architecture to be profiled across heterogeneous cell populations, but the resulting contact maps are extremely sparse — particularly at the 25 kilobase resolution considered biologically meaningful for TAD boundary detection. This sparsity has limited the reliability of TAD reorganisation calls from scHi-C data.

DiffDomain-Spectrum addresses this by applying a spectral decomposition approach to the aggregated contact maps, exploiting the mathematical structure of chromatin contact data to improve signal extraction at low coverage. The framework is designed to compare reorganised TADs between conditions or cell types, rather than simply calling TAD boundaries within a single sample.

For researchers in 3D genomics, chromatin biology, cancer genomics, and developmental genetics, improved detection of TAD reorganisation in single-cell data is a methodologically important advance. The preprint has not yet undergone peer review. Educators teaching genome organisation and chromosome biology may find the tool conceptually relevant to discussions of how chromatin architecture varies between cell states.