CryoLVM: Self-supervised Learning from Cryo-EM Density Maps with Large Vision Models

Weining Fu, Kai Shu, Qiangfeng Cliff Zhang · 2026

Read CryoLVM: Self-supervised Learning from Cryo-EM Density Maps with Large Vision Models by Weining Fu, Kai Shu, Qiangfeng Cliff Zhang (2026) 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-2602-02620.txt.

Description

Weining Fu and 3 co-authors build a foundation model for cryo-electron microscopy density maps, in a 2026 arXiv paper. Cryo-EM reveals biomolecular assemblies at near-atomic resolution, and data volume and complexity are growing fast. Deep learning models built for one task scale poorly across the many analyses researchers need.

CryoLVM learns structural representations directly from measured density maps paired with solved structures. A joint-embedding predictive design, the same family used in self-supervised vision, pairs with a specialized backbone network. A novel histogram-based distribution alignment loss speeds convergence and sharpens fine-tuning. The team tests three downstream tasks: density map sharpening, super-resolution, and missing wedge restoration, an artifact of limited tilt angles in the microscope. CryoLVM beat state-of-the-art rivals across multiple quality metrics on all three. The authors present it as a flexible base for the wider cryo-EM workflow.

Catalog record

CREATORWeining Fu, Kai Shu, Qiangfeng Cliff Zhang
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE57 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2602-02620

Browse the full library · About the Tower of Babel