Related Experiment Video
Updated: Aug 6, 2026
06:41
Enhancing Density Maps by Removing the Majority of Particles in Single Particle Cryogenic Electron Microscopy Final Stacks
Published on: May 10, 2024
AI tools for cryo-EM: Protein particle picking, density map enhancement, and atomic model building
Ashwin Dhakal1, Rajan Gyawali1, Joel Selvaraj1
1Department of Electrical Engineering and Computer Science, University of Missouri, Columbia, MO, United States; NextGen Precision Health, University of Missouri, Columbia, MO, United States.
Progress in Molecular Biology and Translational Science
|July 16, 2026
Summary
AI tools automate protein structure determination using cryo-electron microscopy (cryo-EM). These deep learning models enhance particle picking, density map quality, and atomic model building for faster, more accurate results.
Area of Science:
- Structural Biology
- Computational Biology
- Biophysics
Background:
- Single-particle cryo-electron microscopy (cryo-EM) workflows are transitioning from manual to automated processes.
- Artificial intelligence (AI) and deep learning are key drivers of this automation in protein structure determination.
Purpose of the Study:
- To present an end-to-end suite of deep learning-based AI tools for cryo-EM data analysis.
- To address particle picking, density map enhancement, and atomic model building stages.
Main Methods:
- CryoTransformer and CryoSegNet for AI-based particle picking using advanced neural network architectures.
- CryoTEN, a transformer-based model for enhancing cryo-EM density maps.
- Cryo2Struct and MICA for automated de novo and multimodal atomic model building, integrating with AlphaFold3 predictions.
Main Results:
- AI particle pickers achieve high precision-recall, improving 3D reconstructions and outperforming existing methods.
- CryoTEN enhances density map interpretability and downstream modeling over 10x faster than alternatives.
- Cryo2Struct and MICA generate more complete and accurate atomic models, with MICA achieving near-experimental accuracy.
Conclusions:
- The developed AI tools significantly advance automated protein structure determination via cryo-EM.
- Future directions include addressing open challenges in cryo-EM data analysis and further AI integration.
