HiC-ECC: Hi-C Enhancement, Comparison, and Calling Pipeline

Overview

HiC-ECC is an end-to-end pipeline for Hi-C data analysis, integrating resolution enhancement, pairwise comparison, and structure calling into a single config-driven workflow. It is designed to be accessible to biologists with minimal computational experience, while remaining flexible for advanced users.

The pipeline consists of three modules:

1. Enhance — Improve Hi-C resolution using deep learning (DeepHiC)
2. Compare — Pairwise comparison of chromatin contact maps and unique region extraction (CHESS)
3. Call — Loop and TAD calling on enhanced maps (Chromosight, hicFindTADs)

Installation

# Clone the repository
git clone https://github.com/OroujiLab/HiC-ECC
cd HiC-ECC

# Create and activate the conda environment
conda env create -f environment.yml
conda activate hic

Quick Start

1. Edit config/config.yaml with your paths and samples
2. Run the full pipeline:

python run.py --config config/config.yaml --module all

Or run individual modules:

python run.py --config config/config.yaml --module enhance
python run.py --config config/config.yaml --module compare
python run.py --config config/config.yaml --module call

For SLURM clusters, parallelize the compare module across tissues:

for i in {0..7}; do sbatch submit_compare.sh $i; done

Repository Structure

HiC-ECC/
├── src/                  # Core Python modules
│   ├── enhance.py
│   ├── compare.py
│   └── call.py
├── notebooks/
│   └── tutorials/        # Step-by-step Jupyter notebooks
│       ├── enhance.ipynb
│       ├── compare.ipynb
│       ├── call.ipynb
│       └── visualize.ipynb
├── config/
│   └── config.yaml       # All pipeline parameters
├── data_example/         # Small example dataset for testing
├── results_example/      # Expected outputs
├── figures/              # Pipeline figures
├── run.py                # CLI entry point
└── environment.yml       # Conda environment

Configuration

All parameters are set in config/config.yaml. Key fields:

samples:
  - Brain
  - Kidney

genome: mm10
resolution: 10000
hicpro_dir: /path/to/hicpro_output
output_dir: /path/to/output

enhancement:
  method: deephic
  checkpoint: /path/to/deephic_raw_16.pth

comparison:
  window: 310000
  step: 10000

calling:
  loops:
    tool: chromosight
  tad:
    tool: hicFindTADs

Dependencies (conda env: hic, Python 3.9)

• cooler
• cooltools
• pairtools
• chess-hic
• fithic
• tadbit
• cooler-ontad
• coolpuppy
• hicrep
• hic-straw
• hicexplorer
• PyTorch 2.1.1
• torchvision
• fanc
• matplotlib
• seaborn
• plotly
• tadtool
• bedtools
• samtools
• deeptools
• macs2
• pybedtools
• pybigwig
• bioframe
• numpy
• scipy
• pandas
• scikit-learn
• numba
• dask

Full environment: environment.yml

Methods and Tools

Pre-processing

• HiC-Pro — https://github.com/nservant/HiC-Pro

Enhancement

• DeepHiC — https://github.com/Jakob-Zerbs/DeepHiC/tree/dev
• HiCplus — https://github.com/Jakob-Zerbs/hicplus
• DeepLoop — https://github.com/Jakob-Zerbs/DeepLoop

Comparison

• CHESS — https://github.com/Jakob-Zerbs/chess

Calling

• hicFindTADs — https://github.com/deeptools/HiCExplorer
• Chromosight — https://github.com/koszullab/chromosight

Visualization

• FAN-C — https://github.com/vaquerizaslab/fanc
• HiCPlotter — https://github.com/akdemirlab/HiCPlotter

Tested Samples

Pipeline validated on mm10 mouse tissues: Kidney, Spleen, Liver, Large Intestine, Small Intestine, Lung, Pancreas, Brain.