Advanced algorithms to reveal continuous and discrete structural dynamics in cryo-EM data
3D Flexible Refinement
3DFlex introduces a physically meaningful model of continuous heterogeneity by learning a flexible deformation field that maps particle-specific motion. Through iterative optimization of neural and cryo-EM parameters, 3DFlex reconstructs an accurate reference volume representing both static and dynamic structural components, enabling visualization of subtle motions that bridge conformational states.
3D Variability Analysis
3DVA characterizes variability through linear subspace modeling, identifying major axes of motion across a dataset. By projecting particles along these modes, researchers can visualize molecular transitions, quantify flexibility, and generate ensemble representations of structural dynamics. 3DVA serves as a foundation for further analysis with 3DFlex or focused refinement techniques.