DeTOKI identifies and characterizes the dynamics of chromatin topologically associating domains in a single cell

Abstract

The human genome has a dynamic, well-organized hierarchical 3D architecture, including megabase-sized topologically associating domains (TAD). TADs are a key structure of the genome regulating nuclear processes, such as gene expression, DNA replication and damage repair. However, owing to a lack of proper computational tools, TADs have still not been systematically and reliably surveyed in single cells. In the present work, we developed a new algorithm to decode TAD boundaries that keep chromatin interaction insulated (deTOKI) from ultra-sparse Hi-C data. By nonnegative matrix factorization, this novel algorithm seeks out for regions that insulate the genome into blocks with minimal chance of clustering. We found that deTOKI outperformed competing tools and that it reliably identified TADs with single-cell Hi-C (scHi-C) data. By applying deTOKI, we found that domain structures are prevalent in single cells. Further, although domain structures are highly dynamic between cells, TADs adhere to the ensemble, suggesting tight regulation of single-cell TADs. Finally, we found that the insulation properties of TAD boundaries have major effect on the epigenetic landscape in individual cells. In sum, deTOKI serves as a powerful tool for profiling TADs in single cells.

Single-cell Hi-C data analysis: safety in numbers

“…Aggregation confirms the presence of these chromatin features in individual cells [ 32 ], and there is specialised software for this purpose [ 29 ]. With features calling , the positions of individual loops [ 82 ], TADs [ 28 , 60 , 75 , 93 ] and compartments [ 75 , 86 ] are found directly in the scHi-C map, demanding high-quality scHi-C maps and providing insight into variability between individual cells. For example, the positions of TADs in individual cells demonstrated higher stability of TAD boundaries between individual cells of Drosophila than between mouse oocytes [ 93 ].…”

“…Nagano et al [ 69 ] observe the cell cycle-dependent embedding of scHi-C by calculating the pairwise symmetric Kullback–Liebler divergence on vectors of distance decays and subsequent spectral embedding. Collombet et al [ 16 ] apply uniform manifold approximation and projection (UMAP) to vectors of TAD contact profiles; Li et al [ 60 ] perform PCA on pairwise similarities of TAD profiles; Tan et al [ 87 ] calculate the compartment score profiles for each cell, take 20 principal components and then visualise it with t-distributed stochastic neighbour embedding (t-SNE) . One of the most generalised approaches is HiCRep [ 100 ], which calculates a similarity matrix between each pair of individual cells, taking the stratum-adjusted correlation coefficient (SCC) measure of similarity.…”