Nuc3DMap delineates finer nucleosome-based three-dimensional chromatin architecture

Abstract

Currently, there lacks a method to delineate finer nucleosome-based 3D chromatin architecture. In this study, we develop a novel computational method, nucleosome-based 3D chromatin map (Nuc3DMap), to integrate MNase-seq, Hi-C and Micro-C data. Nuc3DMap implements an ensembled LOESS-sparse Knight-Ruiz algorithm to efficiently balance nucleosome-based contact matrix and adapts a bottom-up approach to identify nano-scale chromatin organization including nucleosome-based topological domain (NucD), boundary (NucB), gap (NucG) as well as nucleosome-based chromatin interaction loci (NucIL) and chromatin loop (NucL). We evaluate Nuc3DMap on publicly available data in H1 and GM12878 cells and identify the distinct subtypes of functional NucBs and NucLs, respectively, by integrating them with more than a hundred of multi-omics modalities data. Interestingly, functional NucBs exert distinct nucleosome organizational features and different insulation capability; and Pol2-related NucLs implicated Pol2 as a noncanonical regulator in mediating gene looping and cell type specificity. Furthermore, our work systematically demonstrates the relationship between interaction intensity and transcriptional activity, unveiling an activation-like behavior of chromatin looping. Overall, Nuc3DMap represents a conceptual advancement in chromatin conformation analysis, offering a refined view of genome organization from the fundamental nucleosome level to higher-order chromatin structures, as well as serves as a foundational framework for advancing computational methods development in dissecting nano-scale 3D genome organization, enabling new discoveries and fostering innovation in chromatin organization modeling.