cryoDRGN

cryoDRGN reconstructs continuous distributions of 3D electron density maps from cryo-electron microscopy single-particle data to characterize per-particle conformational and compositional heterogeneity.

Key Features:

• Deep Neural Network Integration: Leverages deep neural networks to model complex cryo-EM single-particle data and structural variability.
• Continuous Density Map Reconstruction: Directly reconstructs continuous distributions of 3D density maps instead of discrete classes.
• Per-Particle Heterogeneity Mapping: Maps heterogeneity at the level of individual particles to identify structural variations across a dataset.
• Exploratory Analysis Capabilities: Enables generation of density maps and extraction of specific particle subsets for downstream analysis with other tools.
• Trajectory Generation: Produces trajectories that visualize large-scale molecular motions across the reconstructed continuum.

Scientific Applications:

• Residual heterogeneity detection: Uncovers residual heterogeneity in high-resolution cryo-EM datasets such as the 80S ribosome and the RAG complex.
• State discovery during assembly: Reveals new structural states observed in assembling 50S ribosomes.
• Visualization of continuous motions: Visualizes extensive continuous conformational motions in dynamic complexes such as the spliceosome.

Methodology:

Uses deep neural networks to directly reconstruct continuous distributions of 3D density maps from cryo-EM single-particle data and to map per-particle heterogeneity.