Cryo-EM images are intrinsically low dimensional
Luke Evans, Octavian-Vlad Murad, Lars Dingeldein, Pilar Cossio, Roberto Covino, Marina Meila · 2025
Read Cryo-EM images are intrinsically low dimensional by Luke Evans, Octavian-Vlad Murad, Lars Dingeldein, Pilar Cossio, Roberto Covino, Marina Meila (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-11249.txt.
Description
Luke Evans and co-authors show that cryo-electron microscopy images hide a simple inner structure in a 2025 paper. Cryo-EM photographs frozen molecules from many angles, and neural methods now infer molecular shapes from learned internal representations. What those representations look like geometrically was unclear.
The team applied manifold learning to representations of the hemagglutinin protein, from both simulated and real data. The high-dimensional codes turned out to lie on smooth, low-dimensional surfaces. Simulated data covered the real data's surface well. Reading the surface's main directions linked them to physical parameters of the molecule. A simple hidden geometry strengthens trust in these inference tools and may guide better ones.
Catalog record
| CREATOR | Luke Evans, Octavian-Vlad Murad, Lars Dingeldein, Pilar Cossio, Roberto Covino, Marina Meila |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 42 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2504-11249