Description |
We build a stochastic folding model, which propose that the inherent stochastic folding of heterogeneous fluidlike chromatin fibers within a crowded nucleus underlies the single-cell domain formation and the emergence of population-averaged TADs. By only using DNA-accessibility data to parameterize the DNA-packing density along the chromatin fiber, we model a heterogeneous chromatin fiber as a heteropolymer and simulate random folding of the heteropolymer under a confined space. Our stochastic folding model excludes any directed process or interaction, including cohesin-mediated chromatin extrusion, attraction of homotypic chromatin to each other (i.e. phase separation), and attraction of heterochromatin to the nuclear lamina. We find that the stochastic folding model can simultaneously recapitulate the features of both the single-cell domains and the population-averaged TADs. Furthermore, the stochastic folding model can quantitatively reproduce the genomic positions of the boundaries of TADs seen in ensemble-averaged Hi-C contact maps and FISH spatial-distance matrices. More importantly, the stochastic folding model allows, for the first time, de novo prediction of the dynamic changes in chromatin organization during cell differentiation, by using chromatin accessibility data as the only input. These results reveal that it is the two universal physical properties of 1D chromatin fibers: chromatin fluidlike behavior and heterogeneity in DNA-packing density along chromatin, that play a vital role in the 3D genome organization at the kilobase-to-megabase scale. |