Fig. 6: Identification of A/B compartments and TADs based on the completed single-cell contact maps.

a Compartment identification in chromosome 19 for 351 mESC single cells at 30kb-resolution. PC1 scores based on the bulk Hi-C contact maps, the pooled single-cell contact maps, and all 351 single cells are shown separately. (b, c) Regions with higher TAD boundary scores tend to have b higher CTCF binding ChIP-seq signal strength (the number of called TAD boundaries n = 789, one-sided Mann-Whitney U test p-value = 2.6 × 10−4) and c higher number of CTCF peaks (High: <20% quantile; Mid: 20%–80%; Low: >80%; the number of called TAD boundaries n = 789, one-sided Mann-Whitney U test p-value = 2.7 × 10−2). For b, c, the center lines of boxplots show the median; the upper and lower box limits show the 25th and 75th percentiles, respectively. The whiskers extend up to 1.5 times the interquartile range away from the limits of the boxes. d Example of TAD boundary identification at chr19:13,620,000-15,750,000. Seven out of eight predicted TAD boundaries are supported by the bulk CTCF ChIP-seq peaks. The TAD boundary without CTCF peak contains a SINE B2 transposable element with a B-box for TFIIIC. e, f Tensor-FLAMINGO’s single-cell predictions reveal the patterns of structural variabilities across different genomic regions. The RMSD is calculated between the 3D chromatin structures of every single cell and the averaged structures of all cells. Lower RMSD indicates less structural variability and the specific genomic region is more structurally stable across single cells. e Compartment A regions show lower RMSD compared with compartment B regions (the number of called chromatin compartments n = 121, one-sided Mann-Whitney U test p-value:1.8 × 10−3). f Genes with GM12878-specific expressions show lower RMSD compared with other genes (the total number of genes n = 19,003, one-sided Mann-Whitney U test p-value:0.037). For e, f, the center lines of boxplots show the median; the upper and lower box limits show the 25th and 75th percentiles, respectively. The whiskers extend up to 1.5 times the interquartile range away from the limits of the boxes. Source data are provided as a Source Data file.