The repressive genome compartment is established early in the cell cycle before forming the lamina associated domains II (re-analysis)

Three-dimensional (3D) genome organization is thought to be important for regulation of gene expression. Chromosome conformation capture-based studies have uncovered ensemble organizational principles such as active (A) and inactive (B) compartmentalization. In addition, large inactive regions of the genome associate with the nuclear lamina, the Lamina Associated Domains (LADs). Here we investigate the dynamic relationship between A/B-compartment organization and the 3D organization of LADs. Using refined algorithms to identify active (A) and inactive (B) compartments from Hi-C data and to define LADs from DamID, we confirm that the LADs correspond to the B-compartment. Using specialized chromosome conformation paints, we show that LAD and A/B-compartment organization are dependent upon chromatin state and A-type lamins. By integrating single-cell Hi-C data with live cell imaging and chromosome conformation paints, we demonstrate that self-organization of the B-compartment within a chromosome is an early event post-mitosis and occurs prior to organization of these domains to the nuclear lamina. Hi-C data were mapped using BWA mem followed by processing, normalization, and analysis using HiFive. DamID data were segmented using LADetector. ------------------------------ Hi-C data were aligned using BWA mem. Hi-C data paired ends were aligned independently. All reads were mapped using default parameters. Hi-C data were analyzed using HiFive v1.4. Fragments were filtered to include only those with at least one interaction. Counts were normalized using the Binning algorithm, normalizing for GC content, fragment length, and mappability using 20, 20, and 10 bins, respectively. GC content and fragment length parameters were split into evenly sized bins and optimized while mappability was split into fixed bin sizes and seeded but not optimized. Seed compartment states were generated using a bin size of 10 kb, dynamically binning with a minimum read count of 3 and finding the sign of the first eigenvector of the log of the observed over expected bin values. Likelihood compartment scores were calculated using a burnin period of 20 iterations and going to a maximum of 200 iterations. Only interactions of 500 Kb or greater were used and fragments were filtered to have at least five interactions greater than 500 Kb. The minimum reads for modeling the distance curve was 10,000 and each iteration the top scoring 50% of state-changing bins were updated. ES LAD calls were found using the original analysis log2 ratio scores. LADs were determined using the LADetector (https://github.com/thereddylab/pyLAD) segmentation module with default parameters. DamID-array data were processed using LADetevtor v1 (Harr et al., doi: 10.1083/jcb.201405110) where they were quantile normalized (preprocessCore R package), smoothed and segmented (DNAcopy R package). Segmentation was consolidated, further refined and LADs were returned. LADs from individual arrays were merged. non-LADs were defined as complement regions to LADs. mm8, mm9 BigWig files were generated using the UCSC bedGraphToBigWig converter tool. Compartment scores were created using HiFive. Bed file was created using LADetector. .LADs and .nonLADs bed files were defined by LADetector v1 and complement nonLAD regions were defined by Galaxy tool "Complement intervals of a dataset". --------------------------------- GSM2026269, GSM202670, GSM2026271: MEF_mm9_boundary_10K.bw GSM426758: mm9_LaminB1.bed GSM1372713, GSM1372714: MEF_Chr11_Chr12_mm8.LADs.txt, MEF_Chr11_Chr12_mm8.nonLADs.txt