The abstract is read at the publisher; the summary is JClub's.
An AI-enabled, vesicle geometry-guided reconstruction from untilted data achieves 2.7 Å for influenza A virus hemagglutinin and 3.1 Å for HCoV-OC43 prefusion spike.
- Why it matters: 3D reconstruction of membrane proteins in native lipid environments is still constrained by tradeoffs between efficiency and resolution in current single-particle and tomographic workflows.
- What they did: The isSPA-vesicle strategy trains an AI model to fit ellipsoidal membrane geometry, estimate membrane normals, then post hoc filter isSPA-picked particles by orientation–normal angular difference and exclude particles outside vesicles.
- The result: It enables efficient near-atomic reconstructions of membrane proteins from 130 kDa to 2 MDa across two enveloped viruses and three synthetic liposome types, reaching 2.7 Å and 3.1 Å for the specified targets.
The findingWhy it mattersWhat they didThe result
AI summary of the abstract · may contain errors · not medical advice · read the paper.