Abstract:While advances in single-particle cryoEM have enabled the structural determination of macromolecular complexes at atomic resolution, particle orientation bias (the so-called “preferred” orientation problem) remains a complication for most specimens. Existing solutions have relied on biochemical and physical strategies applied to the specimen and are often complex and challenging. Here, we develop spIsoNet, an end-to-end self-supervised deep-learning-based software to address the preferred orientation problem. Using preferred-orientation views to recover molecular information in under-sampled views, spIsoNet improves both angular isotropy and particle alignment accuracy during 3D reconstruction. We demonstrate spIsoNet’s capability of generating near-isotropic reconstructions from representative biological systems with limited views, including ribosomes, β-galactosidases, and a previously intractable hemagglutinin trimer dataset. spIsoNet can also be generalized to improve map isotropy and particle alignment of preferentially oriented molecules in subtomogram averaging. Therefore, without additional specimen-preparation procedures, spIsoNet provides a general computational solution to the preferred orientation problem.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is the preferred orientation problem, which is widespread in single - particle cryo - electron microscopy (cryoEM). This problem leads to an uneven orientation distribution during the three - dimensional reconstruction process, which in turn affects the quality and accuracy of the reconstruction. Especially when dealing with macromolecular complexes with low symmetry or asymmetry, it may lead to defects such as distortion in the reconstructed map, broken secondary structures, and distorted side - chain or nucleotide densities. Existing solutions usually rely on biochemical or physical strategies, which are often complex and challenging and sometimes damage the sample quality. To solve this problem, the paper introduces a self - supervised deep - learning - based method - the single - particle isotropic network (spIsoNet), aiming to restore the angular isotropy of samples showing orientation preference. spIsoNet solves the main problems caused by orientation preference through two modules: (1) the Map Anisotropy Correction module and (2) the Particle Misalignment Correction module. These two modules can be used separately or in combination to overcome the orientation preference problem, improve particle alignment accuracy and angular isotropy, and thus improve the quality of 3D reconstruction. By being applied to a variety of representative biological systems, such as ribosomes, β - galactosidase, and the previously difficult - to - handle hemagglutinin trimer data set, spIsoNet has demonstrated its ability to generate approximately isotropic reconstructions. In addition, spIsoNet can also be generalized to improve map isotropy and particle alignment in sub - tomogram averaging. Therefore, without additional sample preparation procedures, spIsoNet provides a general computational solution to deal with the orientation preference problem.

Overcoming the preferred orientation problem in cryoEM with self-supervised deep-learning

Overcoming the preferred-orientation problem in cryo-EM with self-supervised deep learning

Addressing preferred orientation in single-particle cryo-EM through AI-generated auxiliary particles

Isotropic reconstruction for electron tomography with deep learning

CryoPROS: Correcting Misalignment Caused by Preferred Orientation Using AI-generated Auxiliary Particles

DeepOrientation: Deep Orientation Estimation of Macromolecules in Cryo-electron tomography

A Novel Cryo-Electron Microscopy Support Film Based on 2D Crystal of HFBI Protein

Cryo-forum: A framework for orientation recovery with uncertainty measure with the application in cryo-EM image analysis

Correction of preferred orientation-induced distortion in cryo-electron microscopy maps

SPREAD: A Fully Automated Toolkit for Single-Particle Cryogenic Electron Microscopy Data 3D Reconstruction with Image-Network-Aided Orientation Assignment

End-to-End Simultaneous Learning of Single-particle Orientation and 3D Map Reconstruction from Cryo-electron Microscopy Data

CryoSPIN: Improving Ab-Initio Cryo-EM Reconstruction with Semi-Amortized Pose Inference

A strategy combining denoising and cryo-EM single particle analysis

High-resolution real-space reconstruction of cryo-EM structures using a neural field network

Protein complex structure modeling by cross-modal alignment between cryo-EM maps and protein sequences

Bayesian Perspective for Orientation Estimation in Cryo-EM and Cryo-ET

Orientation Determination of Cryo-EM Projection Images Using Reliable Common Lines and Spherical Embeddings

goCTF: Geometrically optimized CTF determination for single-particle cryo-EM

A deep convolutional neural network approach to single-particle recognition in cryo-electron microscopy

cryoSPHERE: Single-particle heterogeneous reconstruction from cryo EM

Cryo-ZSSR: multiple-image super-resolution based on deep internal learning