# HiCMamba: Enhancing Hi-C resolution and identifying 3D genome structures with state space modeling

## TL;DR

HiCMamba is a deep learning method that improves the resolution of Hi-C data, helping to better understand 3D genome structures with less computational cost.

## Contribution

HiCMamba introduces a novel state space model-based deep learning framework for Hi-C resolution enhancement.

## Key findings

- HiCMamba outperforms existing methods in enhancing Hi-C contact maps.
- The method improves identification of TADs and chromatin loops in 3D genome structures.
- HiCMamba reduces computational resource requirements while maintaining accuracy.

## Abstract

Hi-C technology measures genome-wide interaction frequencies, providing a powerful tool for studying the 3D genomic structure within the nucleus. However, high sequencing costs and technical challenges often result in Hi-C data with limited coverage, leading to imprecise estimates of chromatin interaction frequencies. To address this issue, we present a novel deep learning-based method HiCMamba to enhance the resolution of Hi-C contact maps using a state space model. We adopt the UNet-based auto-encoder architecture to stack the proposed holistic scan block, enabling the perception of both global and local receptive fields at multiple scales. Experimental results demonstrate that HiCMamba outperforms state-of-the-art methods while significantly reducing computational resources. Furthermore, the 3D genome structures, including topologically associating domains (TADs) and loops, identified in the contact maps recovered by HiCMamba are validated through associated epigenomic features. Our work demonstrates the potential of a state space model as foundational frameworks in the field of Hi-C resolution enhancement. The data and source code used in this work are available at GitHub: https://github.com/myang998/HiCMamba.

Understanding the 3D structure of chromosomes within the cell nucleus is fundamental to deciphering gene regulation. A key technology known as Hi-C allows us to map this 3D genome, but obtaining high-resolution data is often hindered by high costs and technical challenges. This results in the prevalence of low-resolution data, which obscures the fine-scale structural details essential for analysis. To address this limitation, we have developed HiCMamba, a novel deep-learning framework that, for the first time, leverages a state space model for this task. In this study, we introduce HiCMamba to computationally enhance low-resolution Hi-C contact maps to high-resolution quality. Our method is designed to effectively capture both the long-range and local chromatin interactions that define 3D genome architecture. Experimental results demonstrate that HiCMamba outperforms existing state-of-the-art methods, achieving superior accuracy while significantly reducing computational resource requirements. We show that contact maps enhanced by HiCMamba lead to more precise identification of critical genomic structures, including TADs and chromatin loops. Our work provides researchers with a more powerful and accessible tool for high-resolution 3D genome analysis. These advancements not only facilitate a deeper understanding of the interplay between genome structure and cell-specific gene regulation but also establish the potential of state space models as a foundational framework for future innovations in genomics research.