What is LoopSage?¶
LoopSage is a program that turns a list of chromatin loops into an ensemble of 3D genome structures. It models how the protein complex cohesin pushes DNA through itself and builds loops, and it checks the result against your data.
The idea in one minute¶
Inside the cell nucleus, DNA is folded. A key part of the folding is loop extrusion: a ring-shaped protein complex (cohesin) lands on the chromatin fibre and slides along it, pulling in DNA and growing a loop. It stops at roadblocks, mainly CTCF proteins bound to specific DNA sites, which are called anchors of the loop.
LoopSage reproduces this process in a computer:
- Input. You give the loops measured in an experiment (for example ChIA-PET, HiChIP or Hi-C loops) and a genomic region.
- Loop extrusion model. Each cohesin is a loop-extruding factor (LEF), a pair of legs on the chromatin fibre. A Monte Carlo algorithm moves the legs at random and keeps moves that make the loops agree better with your data. After many steps it has produced many plausible loop configurations, an ensemble.
- 3D structures. A short molecular dynamics simulation turns every loop configuration into a 3D polymer structure.
- Checks. The simulated contacts are compared with the experiment (loop strength, contact decay with distance, insulation score, compartments).
- Report. Everything is collected in one HTML report with figures, tables and plain-language explanations.
Words used in these notes¶
| Term | Meaning |
|---|---|
| Bead | One piece of the chromatin fibre in the model, covering a fixed stretch of DNA (for example 2 kb). The region is a chain of beads. |
| Loop | A pair of beads (left leg, right leg) held together by a LEF. |
| Anchor | A bead where loops tend to start or stop, typically a CTCF site. |
| LEF | Loop-extruding factor, the model of one cohesin complex. |
| Monte Carlo (MC) | A method that explores many configurations by random moves, accepting or rejecting each move with a probability set by an energy. |
| Energy | A number that is low when the loops agree with the data and with the physics of the model. Low-energy configurations are more likely. |
| MD | Molecular dynamics, a physical simulation of the 3D polymer. |
| PET | Paired-end tag, one sequencing read pair that supports a loop. More PET counts means a more reliable loop. |
| BEDPE | A text file format with one loop per line: two genomic intervals (the two anchors) plus a score. |
What you give and what you get¶
Input
- A loop file:
.bedpe(loops), or.bed/.narrowPeak(CTCF sites). - A region, for example
chr1:77000000-95000000. - Optional: CTCF orientation (inside the BEDPE file, or read from a reference genome by LoopSage), BigWig tracks, a compartment track.
Output (in one folder per run)
report/report.html, the full report with figures and explanations.visualizations/, every figure as PNG and SVG.structures/, 3D structures (.cif), an interactive 3D viewer and videos.data/, the numerical results; maps are saved as.npyfiles indata/metadata/.
Quick start¶
loopsage -f loops.bedpe -r chr1:77000000-95000000
Then open results/chr1_77000000_95000000_1/report/report.html in a browser.
Install the program first: see Installation. All options are listed in the Reference.
Citation¶
If you use LoopSage, please cite:
- Korsak, S. and Plewczynski, D. LoopSage: An energy-based Monte Carlo approach for the loop extrusion modelling of chromatin. Methods (2024).
- Jodkowska, K., Parteka-Tojek, Z., Agarwal, A., Denkiewicz, M., Korsak, S., Chiliński, M., Banecki, K., and Plewczynski, D. Improved cohesin HiChIP protocol and bioinformatic analysis for robust detection of chromatin loops and stripes. bioRxiv (2024). doi:10.1101/2024.05.16.594268