Reconstruction of Diploid High-Order 3D Genome Interactions from Long Noisy Concatemers

Differential high-order chromatin interactions between homologous chromosomes play pivotal roles in many biological processes. However, their elucidation has been hindered by technical difficulties. Traditional 3C methods mainly expose 2-way interactions and offer limited haplotype information. In response, we addressed challenges in harnessing merely nanopore high-order concatemer (Pore-C) reads to delineate diploid high-order chromatin interactions. By training a cutting-edge deep learning model and making statistical analysis, we achieved superior SNV calling and haplo-tagging for noisy short monomers. Learning the haplotype characteristics of high-order concatemers allowed us to devise a progressive haplotype imputation strategy, which elevated the haplotype informative Pore-C contact rate by 14.1-fold to 76% in the HG001 cell line, eclipsing Hi-C's rate by over ten times. Overall, our diploid 3D genome interactions surpassed conventional methods in noise reduction and contact distribution uniformity, with over one magnitude advantages in haplotype informative contact density and genomic coverage rate. Dip3D enabled the unveiling of haplotype high-order interactions in many previously overlooked genomic regions and the investigation of their relationship with allele-specific expression, especially concerning X-chromosome inactivation. Our approach stands out as a robust and accessible solution for the high-quality reconstruction of diploid high-order 3D genome interactions.