Enhancing Hi-C Signals with ChIP-seq Guided Deep Learning for Chromatin Interaction Analysis

High-throughput chromosome conformation capture (Hi-C) data provides key information for understanding 3D genome architecture, although there are a lot of things that affect the quality of the data, including the use of enzymes and noise. Here, we suggest a new deep learning-based model called HiCEnhancer that uses residual blocks, attention mechanisms, and multi-task loss functions to improve the data quality of Hi-C contact maps by integrating Hi-C and ChIP-seq data. The results of the study, which used GM12878 as a case study, showed that the HiCEnhancer performed better than the baseline techniques in terms of the biological consistency of epigenetic signals, as well as the Pearson correlation coefficient, and in terms of Pearson correlation, which suggested improvements in both prediction accuracy and biological relevance. Our model is a potent tool for downstream research as well as chromatin interaction analysis.