Scope and goals

RepliSage follows one chromosome through the whole cell cycle (G1, S, G2/M) in a single model, with S phase as its focus. Most tools treat loop extrusion, compartments or replication separately; here the three can influence each other, which lets you test mechanistic ideas with your own data. The paper calls it a proof-of-concept framework, and that is a good way to use it.

Goals

• Simulate all phases and obtain a 3D ensemble for each, from your own loops (BEDPE) and replication timing.
• Study S phase: how moving forks affect loops, epigenetic states and sister chromatids.
• Test hypotheses: fork speed, origin firing, barrier strength, whether forks organise epigenetic states.
• Model replication stress (RS): slower forks and more origins firing.

The phases

Phase	What the model does
G1	One DNA copy. LEFs (cohesin) extrude loops, epigenetic states follow the timing-derived field
S	Replikator provides fork positions. Forks push LEFs and may organise states. Two chains are tied where DNA is copied and gradually separate
G2/M	Replication ties are off. A second LEF family (condensin, more processive) and random excluded-volume relaxation (topoisomerase) are added; a gentle force separates the sisters

What the paper reports

• Loops get shorter in S phase and shorter still under RS, as in single-cell Hi-C. Cohesin HiChIP under mild RS (aphidicolin, hydroxyurea) shows the same.
• Compartments stay largely stable under RS. The model's epigenetic states correlate with Hi-C compartments (above 80%).
• If forks organise epigenetic states, RS gives more B to A switching.
• The chromosome decompacts in S and compacts in G2/M, as in NucDynamics models from single-cell Hi-C.

Paper settings for chromosome-wide runs: 10,000 beads, 1000 cohesin and 1000 condensin LEFs, \(T_{\text{mc}} = 1.8\), \(2\times10^5\) Monte Carlo steps. The defaults in config.ini are lighter.

For the formalism behind the model, see the Hamiltonian idea. Please read Known limitations before drawing conclusions from a result.