Topology-aware multiscale modeling of viral genomes reveals stability determinants in circoviruses

Understanding how viral genomes are organized under extreme spatial confinement remains a fundamental challenge in structural virology. Icosahedral viruses, despite offering high-resolution capsid structures through cryo-electron microscopy and X-ray crystallography, present a major obstacle: their genomes do not obey icosahedral symmetry and are thus averaged out during standard reconstruction procedures, leaving genome topology largely unresolved. Computational modeling offers a complementary avenue, but existing approaches often rely on simplified polymer representations that fail to capture sequence-specific features and the extreme compaction observed in small DNA viruses. Here, we focus on Porcine Circovirus type 2 (PCV2), a member of the Circoviridae family and one of the smallest autonomous mammalian viruses, which packages a circular ~1.7 kb single-stranded DNA genome into a ~20 nm T=1 icosahedral capsid at one of the highest DNA packing densities found in nature. We introduce an integrative methodology combining AI-based structural prediction, lattice Monte Carlo simulations, and multiscale molecular dynamics to generate and simulate three-dimensional topological models of the complete PCV2 virion. Our results demonstrate that multiple distinct genome arrangements can produce virions with indistinguishable external morphology, yet differ substantially in their internal stress distributions and predicted particle stability. These findings suggest that PCV2 populations comprise energetically heterogeneous assemblies with implications for infectivity, uncoating, and environmental persistence, while providing a generalizable framework for modeling genome topology in other confined viral systems.