A comprehensive foundation model for cryo-EM image processing
Published in Nature Methods, 2026
Abstract
Cryo-IEF establishes a foundation-model paradigm for cryo-EM particle processing by pretraining representations on large-scale particle-image data and transferring them to core downstream tasks. The model supports structural classification, pose-aware clustering, particle-quality ranking, and automated reconstruction, showing how reusable learned particle embeddings can improve automated analysis in single-particle cryo-EM. By connecting representation learning with practical cryo-EM workflows, the work provides a foundation for scalable particle assessment and structure-determination systems.
Recommended citation: Yang Yan, Shiqi Fan, Fajie Yuan, Huaizong Shen. (2026). "A comprehensive foundation model for cryo-EM image processing." Nature Methods, 23(1), 88-95. DOI: 10.1038/s41592-025-02916-8.
Publisher page