![[2602.02620] CryoLVM: Self-supervised Learning from Cryo-EM Density Maps with Large Vision Models](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.02620] CryoLVM: Self-supervised Learning from Cryo-EM Density Maps with Large Vision Models
Summary
CryoLVM introduces a self-supervised learning model for cryo-electron microscopy (cryo-EM) density maps, enhancing structural representation and performance across various tasks.
Why It Matters
As cryo-EM data grows in complexity, CryoLVM offers a scalable solution that improves the analysis of biomolecular structures. This model's versatility can significantly advance research in structural biology and related fields.
Key Takeaways
- CryoLVM utilizes a Joint-Embedding Predictive Architecture for enhanced learning.
- Introduces a novel histogram-based distribution alignment loss for improved convergence.
- Demonstrates superior performance in density map sharpening, super-resolution, and missing wedge restoration.
Quantitative Biology > Quantitative Methods arXiv:2602.02620 (q-bio) [Submitted on 2 Feb 2026 (v1), last revised 24 Feb 2026 (this version, v2)] Title:CryoLVM: Self-supervised Learning from Cryo-EM Density Maps with Large Vision Models Authors:Weining Fu, Kai Shu, Kui Xu, Qiangfeng Cliff Zhang View a PDF of the paper titled CryoLVM: Self-supervised Learning from Cryo-EM Density Maps with Large Vision Models, by Weining Fu and 3 other authors View PDF HTML (experimental) Abstract:Cryo-electron microscopy (cryo-EM) has revolutionized structural biology by enabling near-atomic-level visualization of biomolecular assemblies. However, the exponential growth in cryo-EM data throughput and complexity, coupled with diverse downstream analytical tasks, necessitates unified computational frameworks that transcend current task-specific deep learning approaches with limited scalability and generalizability. We present CryoLVM, a foundation model that learns rich structural representations from experimental density maps with resolved structures by leveraging the Joint-Embedding Predictive Architecture (JEPA) integrated with SCUNet-based backbone, which can be rapidly adapted to various downstream tasks. We further introduce a novel histogram-based distribution alignment loss that accelerates convergence and enhances fine-tuning performance. We demonstrate CryoLVM's effectiveness across three critical cryo-EM tasks: density map sharpening, density map super-resolution, and missing wedge...
