HiCInterpolate: 4D Spatiotemporal Interpolation of Hi-C Data for Genome Architecture Analysis

In this study, we developed HiCInterpolate, a 4D spatiotemporal interpolation architecture that accepts two timestamp Hi-C contact matrices to interpolate intermediate Hi-C contact matrices at high resolution. HiCInterpolate predicts the intermediate Hi-C contact map using a deep learning-based flow predictor, and a feature encoder and decoder architecture similar to U-Net. Moreover, HiCInterpolate is capable of analyzing different 3D genomic structures, including A/B compartments, chromatin loops, TADs, and 3D genome structure through its analysis pipeline. HiCInterpolate achieved strong PSNR, SSIM, GenomeDISCO, HiCRep, LPIPS scores along with biological validation such as chromatin loops, A/B compartments, and TADs. HiCInterpolate demonstrated notable biological reproducibility and established its position as the most advanced computer vision-based interpolation model for intermediate Hi-C contact matrix prediction at high-resolution along with different downstream analyse. AvailabilityHiCInterpolate is publicly available at https://github.com/OluwadareLab/HiCInterpolate.