scHi-C

A Python toolkit for single-cell Hi-C data normalization and analysis

scHi-C is a study project developed at the Bioinformatics Institute for exploring single-cell Hi-C data in Python. It provides a lightweight codebase and an example notebook to illustrate basic workflows in loading, processing and visualizing sparse single-cell Hi-C contact matrices.

Table of Contents

• About Single-Cell Hi-C Data
• Features
• Installation
• Quick Start
• Normalization Methods
• Example Usage
• Data Sources
• References
• Contact

About Single-Cell Hi-C Data

Single-cell Hi-C (scHi-C) captures chromosome conformation in individual cells, revealing cell-to-cell variability in 3D genome organization. Unlike bulk Hi-C, scHi-C data presents unique analytical challenges:

• Ultra-sparse contact matrices (>99.9% zeros)
• Variable sequencing depth across cells
• High technical noise requiring specialized normalization
• Limited statistical power per individual cell

This toolkit implements normalization methods adapted for these challenges, based on established algorithms from the Hi-C analysis ecosystem.

Features

Normalization Methods:

• ICE - Iterative Correction and Eigenvector decomposition (for high-coverage cells)
• VC - Vanilla Coverage normalization
• Coverage-based - Simple coverage normalization for sparse data
• SCALE - Total count scaling (optimal for ultra-sparse data)

Data Processing:

• Automatic method selection based on data sparsity
• Parallel processing for large cell collections
• Memory-efficient sparse matrix operations
• Integration with cooler/cooltools ecosystem

Quality Control:

• Sparsity pattern diagnosis
• Coverage analysis and filtering
• Cell quality assessment
• Normalization validation

Visualization:

• Contact map plotting
• Quality control plots
• Before/after normalization comparisons
• Group-based analysis

Installation

Requirements

• Python 3.8 or higher
• 16GB+ RAM recommended for large datasets

Install from Source

git clone https://github.com/berlogo/scHi-C
cd scHi-C
pip install -e .

Install Dependencies

pip install -r requirements.txt

Verify Installation

# Test basic imports
from schic_normalization import scHiCNormConfig
from schic_class import scHiC
from scHiC_vis import plot_hic_maps
print("scHiCTools imported successfully!")

Quick start

1. Clone the repository

   git clone https://github.com/berlogo/scHi-C.git
   cd scHi-C

2. Open the example notebook

Launch Jupyter and open Example.ipynb to walk through basic data loading, metadata annotation, grouping and visualization steps:

jupyter lab Example.ipynb

Normalization Methods

Method	Best for	Description
SCALE	Ultra-sparse (<10K contacts)	Simple total count scaling
Coverage	Sparse (10K-100K contacts)	Coverage-based normalization
VC	Medium-sparse (100K-1M contacts)	Vanilla Coverage (square root)
ICE	Dense (>1M contacts)	Iterative correction (computationally intensive)

The toolkit automatically selects the optimal method based on your data's sparsity patterns.

Data Sources

This toolkit has been tested on single-cell Hi-C datasets from:

Public Datasets

• GEO: GSE117876 - Human prefrontal cortex (Ramani et al., 2017)

Supported Data Formats

• Cool files (.cool) - Primary format, created by cooler

References

Key Publications

Single-cell Hi-C Methods:

• Tian, D. et al. (2023). Single-cell Hi-C reveals cell-to-cell variability in chromosome structure. Nature Genetics. DOI: 10.1038/s41588-023-01389-1
• Ramani, V. et al. (2017). Massively multiplex single-cell Hi-C. Nature Methods, 14, 263–266. DOI: 10.1038/nmeth.4155
• Nagano, T. et al. (2017). Cell-cycle dynamics of chromosomal organization at single-cell resolution. Nature, 547, 61–67. DOI: 10.1038/nature23001

Hi-C Analysis Tools:

• Abdennur, N. & Mirny, L.A. (2020). Cooler: scalable storage for Hi-C data and other genomically labeled arrays. Bioinformatics, 36, 311-316. DOI: 10.1093/bioinformatics/btz540
• Open2C et al. (2023). Cooltools: enabling high-resolution Hi-C analysis in Python. PLOS Computational Biology. DOI: 10.1371/journal.pcbi.1011214

Normalization Methods:

• Imakaev, M. et al. (2012). Iterative correction of Hi-C data reveals hallmarks of chromosome organization. Nature Methods, 9, 999–1003. DOI: 10.1038/nmeth.2148
• Lieberman-Aiden, E. et al. (2009). Comprehensive mapping of long-range interactions reveals folding principles of the human genome. Science, 326(5950), 289-293. DOI: 10.1126/science.1181369
• Cournac, A. et al. (2012). Normalization of a chromosomal contact map. BMC Genomics, 13, 436. DOI: 10.1186/1471-2164-13-436

Troubleshooting

Common Issues

Import errors

# Make sure your Python files are in the correct location
# and adjust import paths as needed
import sys
sys.path.append('/path/to/your/code/')
from schic_normalization import scHiCNormConfig

"No cells were successfully normalized"

# Try ultra-sparse configuration
config = scHiCNormConfig.for_ultra_sparse_data()
config.method = "SCALE"
# Then run normalize_schic with this config

Memory errors with large datasets

# Process fewer cells at once
config = scHiCNormConfig()
config.chunk_size = 50  # Reduce from default
config.use_float32 = True  # Use less memory

ICE convergence issues

# Use simpler methods for sparse data
config = scHiCNormConfig(method="coverage")  # Instead of ICE