HiC4D-SPOT: Spatiotemporal Outlier Detection Tool for Hi-C Data

This repository contains the code and pretrained models for HiC4D-SPOT, an unsupervised deep-learning framework designed to detect spatiotemporal anomalies in Hi-C data.

---

🔧 Setup Instructions

Before running any script, configure the necessary parameters in args/args_mega.py based on your experiment.

---

📦 Python Environment

• Tested on Python 3.10.14

• Create the conda environment using the provided environment.yml file:

  conda env create -f environment.yml
  conda activate hic4d-spot

• More info: Conda Documentation

---

📂 Datasets

Download and place the datasets in the data directory using the following structure:

Dataset	Description	GEO Accession	DOI	Local Directory
Du	Preimplantation mouse embryos	GSE82185	https://doi.org/10.1038/nature23263	data/data_Du/allValidPairs
Reed	Human cell line dynamics	GSE201376	https://doi.org/10.1016/j.celrep.2022.111567	data/data_Reed/Hi-C
Zhang	Cardiomyocyte differentiation	GSE116862	https://doi.org/10.1038/s41588-019-0479-7	data/data_Zhang/Hi-C and data/data_Zhang/RNA_seq
Cohesin Loss	Auxin-mediated cohesin degradation	GSE104334	https://doi.org/10.1016/j.cell.2017.09.026	data/data_Cohesinloss_Rao/Hi-C

---

⚙️ Data Processing

1. Preprocessing

• Scripts are located in:
  codebase/data_{datasetName}/1_data_generation

• Run the appropriate script based on your dataset.

2. Spatiotemporal Feature Generation

• Script: codebase/src/generate_data.py

• Run:

  python generate_data.py -id mega

---

🧠 Training HiC4D-SPOT

• Script: codebase/src/train.py

• Run:

  python train.py -id mega

---

🔍 Inference

• Script: codebase/src/predict.py

• Run:

  python predict.py -id mega

• Output will be saved in the predictions/ directory.

---

📊 Visualization

• Script: codebase/src/evaluate.py

• Run:

  python evaluate.py -id mega

• Output will be saved inside the eval/ directory within each model’s prediction directory.

---

📖 Citation

If you use HiC4D-SPOT in your research, please cite:

@article{shrestha2025hic4d,
  title={HiC4D-SPOT: a spatiotemporal outlier detection tool for Hi-C data},
  author={Shrestha, Bishal and Wang, Zheng},
  journal={Briefings in Bioinformatics},
  volume={26},
  number={4},
  pages={bbaf341},
  year={2025},
  publisher={Oxford University Press}
}

---

📬 Contact

For questions, suggestions, or issues, please contact:

• Zheng Wang, PhD
• University of Miami
• Email: zheng.wang@miami.edu

---

Thank you for using HiC4D-SPOT!