DynaMight

DynaMight estimates a continuous space of conformations in a cryo-EM dataset by learning three-dimensional deformations of a Gaussian pseudo-atomic model of a consensus structure for every particle image, improving reconstruction of the consensus structure.

Key Features:

• Continuous conformational space estimation: DynaMight learns three-dimensional deformations of a Gaussian pseudo-atomic model to estimate a continuous space of conformations from cryo-EM particle images.
• Improved consensus reconstruction: Inversion of the learned deformations is used to obtain an improved reconstruction of the consensus structure.
• Error estimation through half-set training: Error estimates on the deformations are obtained by independently training two variational autoencoders on half sets of the cryo-EM data.
• Regularization assessment: The tool enables evaluation of regularization of three-dimensional deformations through the use of atomic models and detection of potential model bias artifacts.

Scientific Applications:

• Flexible single-particle cryo-EM analysis: DynaMight addresses continuously flexing molecules in single-particle analysis of proteins from cryogenic-electron microscopy images.
• Experimental cryo-EM dataset processing: The tool has been demonstrated on several experimental cryo-EM datasets for estimating molecular motions and improving reconstructions.

Methodology:

DynaMight learns three-dimensional deformations of a Gaussian pseudo-atomic model of a consensus structure for every particle image, then inverts the learned deformations to obtain an improved reconstruction. Error estimates are obtained by independently training two variational autoencoders on half sets of the cryo-EM data.