GPSeq Reveals the radial organization of chromatin in the cell nucleus

In eukaryotes, a significant portion of the genome consists of lamina-associated domains, which are preferentially located at the periphery of the cell nucleus. However, how exactly the remaining parts of the genome are radially organized in the nucleus remains largely unknown. Here, we describe a method named Genomic loci Positioning by Sequencing (GPSeq), which allows genome-wide measurements of the distance of genomic loci to the nuclear periphery. Using GPSeq, we generated reproducible maps of radial genome organization in human cells, at various resolutions, which revealed radial gradients of genetic and epigenetic features, gene expression, as well as A/B compartments and chromatin loops. We assessed the contribution of various features in predicting radiality, and found GC content to have the strongest predictive power. Finally, we show that by combining Hi-C with GPSeq information to radially position intra- and inter-chromosomal contacts revealed by Hi-C, we are able to make 3D genome structure predictions much more accurately than it was possible thus far. We conclude that GPSeq is able to reveal fundamental aspects of genome architecture. Overall design: 4 GPseq experiments (Exp1-4) were performed in HAP1 cells. 2 experiments (Exp1-2) using HindIII enzyme and the following digestion times: 10 min, 15 min, 30 min, 1 h, 2 h. 2 experiments (Exp3-4) using MboI enzyme and the following digestion times: 1 min, 5 min, 10 min, 30 min. Additional times were performed for Exp1: 45 min and 6 h. Additionally, 2 GPSeq experiments (Exp5-6) were performed on GM06990 cells using HindIII enzyme and the following digestion times: 10 min, 15 min, 30 min, 2 h. Deposited GPSeq scores were calculated using the gpseqc (v2.3.6) pipeline, available on github: https://github.com/ggirelli/gpseqc . "estimated.tsv"; raw GPSeq score (prob_g column) estimated by gpseqc pipeline. "rescaled.tsv"; rescaled GPSeq score (prob_g column) estimated by gpseqc pipeline. All deposited structures were generated starting from Hi-C data available from the 4DN project (accession #: 4DNFI1E6NJQJ); either (*10000.zip and *10000_intra.zip) using only Hi-C data, or (*10000G.zip and *10000G_intra.zip) integrating Hi-C and GPSeq data coming from the first experiment (Exp1) in this series (as briefly described in the chromflock.README file). The chromflock pipeline is available on github: https://github.com/elgw/chromflock