A multi-task graph-clustering approach for chromosome conformation capture data sets identifies conserved modules of chromosomal interactions

Chromosome conformation capture methods are being increasingly used to study three-dimensional genome architecture in multiple cell types and species. An important challenge is to examine changes in three-dimensional architecture across cell types and species. We present Arboretum-Hi-C, a multi-task spectral clustering method, to identify common and context-specific aspects of genome architecture. Compared to standard clustering, Arboretum-Hi-C produced more biologically consistent patterns of conservation. Most clusters are conserved and enriched for either high- or low-activity genomic signals. Most genomic regions diverge between clusters with similar chromatin state except for a few that are associated with lamina-associated domains and open chromatin. ELECTRONIC SUPPLEMENTARY MATERIAL: The online version of this article (doi:10.1186/s13059-016-0962-8) contains supplementary material, which is available to authorized users.