Back

Disentangling polymer confinement from specific-folding interactions reveals the drivers of E. coli chromosome organization

Delafrouz, P.; Nesterova, T.; Harris, F.; Farooq, H.; Du, L.; Liu, J.; Ma, A.; Xiao, J.; Liang, J.

2026-05-23 genomics
10.64898/2026.05.20.726692 bioRxiv
Show abstract

The three-dimensional organization of the bacterial chromosome is critical for gene regulation. Chromosome conformation capture (Hi-C) has enabled genome-wide mapping of chromosomal folding, yet ensemble-averaged contact maps entangle biologically specific-folding interactions (SFIs) with nonspecific polymer compaction and mask single-cell heterogeneity. Here, we developed a polymer-based simulation framework in E. coli to address these limitations. We found that a null model of 300,000 random polymer configurations recapitulated the global chromosomal organization observed in Hi-C data, establishing that most Hi-C signals reflect generic polymer behavior in a confined volume. Contrasting null-model predictions with experimental Hi-C data isolated a small subset of SFIs (< 7%) and generated a specific-fold ensemble of 20,000 single-cell conformations that reproduced chromosomal interaction domains and single-cell heterogeneity. SFIs were enriched in the ter region, reduced nucleoid accessibility, colocalized with cryptic prophages, and depleted in positively supercoiled regions. H-NS and MatP emerged as major chromosome-wide and local determinants of SFIs, respectively. Furthermore, high-SFI regions correlated with stress-adaptive genes, whereas low-SFI regions harbored housekeeping genes. Together, our results established that a small number of biologically encoded SFIs superimposed on a polymer background shape the E. coli chromosome and gene expression, providing a quantitative framework for dissecting chromosome architecture and function.

Matching journals

The top 5 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.