An explicit numerical leveling of non-uniformly distributed sets of 2D projections with the program VUE improves the 3D reconstructions and illustrates sources of image distortion.

Visualizing Structure, Dynamics and Interactions Using Cryo-EM on Simons Foundation

Shin et al. report the cryo-electron microscopy structure and catalytic mechanism of TrhO, a cofactor-independent enzyme that hydroxylates tRNA wobble uridine using molecular oxygen through a reactive...

REliable PIcking by Consensus (REPIC) - an ensemble learning methodology for cryo-EM particle picking - ccameron/REPIC

Micromolar hits with structural validation but therapeutic gaps.

**Abstract:** This paper introduces a novel methodology, Multi-Resolution Graph Neural Networks (MR-GNN), for automated cryo-EM particle alignment and classification, specifically targeting the analysis of dynamic macromolecular conformations. Existing cryo-EM data processing pipelines often struggle with conformational heterogeneity within particle populations. MR-GNN addresses this limitation by representing particle projections as hierarchical graph structures at multiple resolutions, allowing for the simultaneous modeling of global shape and local conformational changes.

We propose a numerical measure to quantify the deviation of a distribution of cryo-EM 2D views from the uniform distribution. This approach allows the traditional two-dimensional projection diagrams to be complemented with quantitative characteristics and simple, easily generated curves that are straightforward to analyze.