A generative network model reveals hierarchical structural memory guiding post-mitotic 3D genome inheritance

How three-dimensional (3D) genome organization is reestablished after mitosis to preserve gene regulatory programs and cell identity remains unresolved. We asked whether mitotic chromosome architecture contributes to cellular memory by preserving structural information while maintaining the flexibility required for cellular adaptation. Here we introduce HiCONA, a generative Bayesian framework that infers hierarchical genome organization and quantifies the structural contribution of individual interactions through an Entropy Impact Score. HiCONA resolves genome organization across multiple scales, revealing structure-defining interaction patterns underlying chromatin architecture, independently of experimental modality. We identify structurally impactful chromatin interactions retained during mitosis delineating distinct trajectories of post-mitotic genome reassembly. These mitotically preserved interactions are enriched for active, transcriptionally competent regulatory elements, consistent with a cis-encoded architectural memory subsequently remodeled by trans-acting regulators. Together, our findings suggest that molecular information retained on mitotic chromosomes guides structural trajectories, providing an architectural basis for epigenetic memory and post-mitotic genome inheritance.