dcHiC: differential compartment analysis of Hi-C datasets

Abstract

Compartmental organization of chromatin and its changes play important roles in distinct biological processes carried out by mammalian genomes. However, differential compartment analyses have been mostly limited to pairwise comparisons and with main focus on only the compartment flips (e.g., A-to-B). Here, we introduce dcHiC, which utilizes quantile normalized compartment scores and a multivariate distance measure to identify significant changes in compartmentalization among multiple contact maps. Evaluating dcHiC on three collections of Hi-C contact maps from mouse neural differentiation (n = 3), mouse hematopoiesis (n = 10) and human LCL cell lines (n = 20), we show its effectiveness and sensitivity in detecting biologically relevant differences, including those validated by orthogonal experiments. Across these experiments, dcHiC reported regions with dynamically regulated genes associated with cell identity, along with correlated changes in chromatin states, replication timing and lamin B1 association. With its efficient implementation, dcHiC not only enables high-resolution compartment analysis but also includes a suite of additional features, including standalone browser visualization, differential interaction identification, and time-series clustering. As such, it is an essential addition to the Hi-C analysis toolbox for the ever-growing number of contact maps being generated. dcHiC is freely available at https://github.com/ay-lab/dcHiC and examples from this paper can be seen at https://ay-lab.github.io/dcHiC.

Rewiring of the 3D genome during acquisition of carboplatin resistance in a triple-negative breast cancer patient-derived xenograft

“…Summarized A-A and B-B compartment strengths were calculated as the mean log2 contact enrichment between the top (A-A) or bottom (B-B) 20% of PC1 percentiles and between the top and bottom 20% of PC1 percentiles for A-B compartment strength, excluding chrY and chrM. The compartmentalization score was calculated as previously described using mean contact enrichments for A-A, B-B, and A-B following this formula: log((A − A) * (B − B)/(A − B) 2 ) 65 .dcHiC (differential compartment analysis of Hi-C) is a method that identifies statistically significant differences in compartmentalization among two or more contact maps, including changes that are not accompanied by a compartment flip 39 . dcHiC first employs a time-and memory-efficient R implementation of singular value decomposition (SVD) to achieve the eigenvalue decomposition of each Hi-C contact map.…”

“…To get a finer insight into the A/B compartment changes, we used dcHi-C which utilizes quantile normalized compartment scores calculated from eigenvector decomposition and a multivariate distance measure to Vol:. ( 1234567890 www.nature.com/scientificreports/ identify statistically significant changes in compartmentalization among multiple contact maps 39 . Furthermore, to avoid considering eigenvectors capturing chromosomal arms, dcHiC checks for eigenvector correlation with gene density and GC content and anticorrelation with chromosome length (Supplementary Table S3).…”