Back

cryoAgent: An agentic workflow for robust and adaptive end-to-end cryo-EM image processing

Li, D.; Yang, S.; Xiao, Q.; Niu, T.; Zhang, Y.; Zhu, Y.; Sun, F.

2026-04-20 biophysics
10.64898/2026.04.16.718662 bioRxiv
Show abstract

Cryo-electron microscopy (cryo-EM) is the mainstream method for structure determination, yet current automated workflows remain rigid and require expert intervention for failure recovery, heterogeneity analysis, and optimization. We present cryoAgent, an agentic workflow for autonomous cryo-EM image processing with adaptive tool use to address these challenges. cryoAgent improves reconstruction quality across diverse datasets, identifies a previously unreported structural state, and outperforms state-of-the-art automated workflows, advancing scalable and discovery-oriented structural biology.

Matching journals

The top 2 journals account for 50% of the predicted probability mass.