Researchers at the European XFEL in Hamburg have directly observed how viral protein shells, or capsids, dynamically deform during dehydration, challenging long-held assumptions about viral rigidity and offering new insights into airborne viral resilience. The study, published in Light: Science & Applications, utilized the facility's SPB/SFX instrument to capture high-resolution X-ray snapshots of the MS2 bacteriophage as it transitioned from a hydrated to a dehydrated state. To simulate natural aerosol transmission, an international team led by researchers from the Max Planck Institute for the Structure and Dynamics of Matter sprayed a liquid viral suspension into a low-humidity chamber. As droplets traveled through the chamber, partial evaporation occurred before intersecting with the X-ray free-electron laser beam. This process generated hundreds of thousands of single-particle diffraction patterns, allowing the team to reconstruct a continuous structural trajectory rather than merely capturing a static end state. The analysis revealed that MS2 capsids do not shrink uniformly. Instead, they undergo asymmetric buckling transitions, deviating significantly from their initial near-perfect icosahedral symmetry. Contrary to the traditional view of viral shells as rigid containers, the findings demonstrate that capsids possess notable mechanical adaptability, with structural changes occurring in localized regions before propagating across the shell. Molecular dynamics simulations identified a flexible protein segment known as the FG loop as the primary trigger for this deformation. As stabilizing water molecules evaporate, the FG loops contract around the capsid three-fold and five-fold pores, pulling the structure into a more compact conformation. Researchers hypothesize this mechanism may reduce genomic exposure and protect viral integrity during drying. A critical enabler of this breakthrough was the integration of single-particle imaging with advanced machine learning. The team employed beta-variational autoencoders to classify structural heterogeneity across the dataset, mapping intermediate conformations that traditional ensemble-averaged methods would have obscured. This approach allowed precise tracking of size and shape variations across a continuous latent space, establishing a new methodological standard for studying dynamic biomolecular processes. The discovery carries significant implications for virology and public health. By elucidating the structural mechanisms that enable viruses to survive environmental stress, the research provides a foundation for developing novel antiviral strategies and airborne pathogen countermeasures. Future investigations will shift toward more physiologically relevant conditions, utilizing saliva-based proxies containing salts and proteins to determine how real-world transmission matrices influence capsid behavior. The analytical framework developed in this study is also expected to extend to other dynamic biological systems, accelerating structural biology research beyond virology.