Sequential-GAM constructs the single-cell geometric 3D genome structure

3D structure of the chromatin is crucial for cell identity and gene regulation. However, how genome-wide hierarchical geometric structure is organized in single cells remains elusive. Here we developed Sequential-GAM (Sequential Genome Architecture Mapping) to construct the hierarchical geometric structure and estimated the radial position of chromosomes, compartments, subcompartments, and genes in single cells by capturing contiguous thin sections of the nucleus. We found that several epigenomic features, including histone modifications and subcompartments, showed radial gradients from the nuclear center to the periphery. Besides, we estimated the variance of hierarchical structure and revealed the dynamic radial position of subcompartments B1 and B2. Furthermore, we defined the quasi-stable topologically associating domains set (q-stable TADs), in which TADs maintain a relatively stable distance to each other. Interestingly, q-stable TADs revealed the correlation between the stability of chromatin structure and transcription activity. Finally, we discovered that the radial distance and the stability of radial positions for genes in single cells were negatively correlated with their expressions. Taken together, Sequential-GAM is able to estimate 3D genome's geometric structure in single cell, and to reveal the structural stability and inter-cell heterogeneity.