Using spIsoNet to address the preferred-orientation problem in cryoEM reconstructions

Area of Science:

• Structural Biology
• Biophysics
• Computational Biology

Background:

• Cryo-electron microscopy (cryoEM) is crucial for determining biological macromolecule structures at high resolution.
• Preferred orientation of specimens on cryoEM grids leads to uneven data sampling and anisotropic reconstruction artifacts.
• These issues can hinder particle alignment and limit the success of near-atomic resolution cryoEM studies.

Purpose of the Study:

• To present a practical protocol for using the self-supervised deep-learning method, spIsoNet, to mitigate preferred orientation problems in cryoEM.
• To offer two workflows: map Anisotropy Correction and particle Misalignment Correction, for improving cryoEM reconstructions.
• To provide guidance for applying spIsoNet to experimental cryoEM data.

Main Methods:

• Implementation of spIsoNet, a self-supervised deep-learning approach.
• Development of two complementary workflows: map Anisotropy Correction and particle Misalignment Correction.
• Integration of spIsoNet with RELION external reconstruction for enhanced particle pose estimation.

Main Results:

• Demonstration of spIsoNet workflows on influenza hemagglutinin (HA) trimer datasets with varying degrees of preferred orientation bias.
• Successful correction of anisotropic artifacts in cryoEM maps.
• Improved particle pose estimation, leading to better cryoEM reconstructions.

Conclusions:

• The spIsoNet software effectively mitigates preferred orientation issues in cryoEM reconstructions.
• The presented workflows provide a robust method for enhancing cryoEM data quality and enabling higher resolution structure determination.
• This protocol offers a valuable resource for researchers facing orientation bias challenges in cryoEM studies.