Deep reconstructing generative networks for visualizing dynamic biomolecules inside cells
Rangan, R.; Khavnekar, S.; Lerer, A.; Johnston, J.; Kelley, R.; Obr, M.; Kotecha, A.; Zhong, E. D.
Show abstract
Advances in cryo-electron tomography (cryo-ET) have produced new opportunities to visualize the structures of dynamic macromolecular machinery in native cellular environments. Here, we describe a machine learning approach that can reconstruct the structural landscape and dynamics of biomolecular complexes present in cryo-ET subtomograms. This method, cryoDRGN-ET, learns a deep generative model of 3D density maps directly from subtomogram tilt series images and can capture states diverse in both composition and conformation. We use this approach to reconstruct the in situ translation dynamics of prokaryotic ribosomes, and we reveal the distribution of functional states during translation elongation populated by S. cerevisiae ribosomes inside cells.
Matching journals
The top 2 journals account for 50% of the predicted probability mass.
Similar papers in this journal
- CryoDRGN-AI: Neural ab initio reconstruction of challenging cryo-EM and cryo-ET datasets 98%
- Learning structural heterogeneity from cryo-electron sub-tomograms with tomoDRGN 97%
- MISO: Microfluidic protein isolation enables single particle cryo-EM structure determination from a single cell colony 96%
Similar papers in this journal
- Beam image-shift accelerated data acquisition for near-atomic resolution single-particle cryo-electron tomography 96%
- DeepETPicker: Fast and accurate 3D particle picking for cryo-electron tomography using weakly supervised deep learning 96%
- Overcoming Resolution Attenuation During Tilted Cryo-EM Data Collection 96%
Similar papers in this journal
- Capturing ribosomal structures in cellular extracts with cryoPRISM: A purification-free cryoEM approach reveals novel structural states 96%
- Cryogenic light microscopy of vitrified samples with Angstrom precision 94%
- Microfluidic protein isolation and sample preparation for high resolution cryo-EM 94%
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.