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

Area of Science:

• Structural Biology
• Biophysics
• Computational Biology

Background:

• Single-particle cryo-electron microscopy (cryo-EM) enables atomic resolution structural determination of macromolecular complexes.
• Particle orientation bias, or the 'preferred orientation' problem, complicates cryo-EM data analysis, leading to map anisotropy and particle misalignment.
• Current solutions involve complex biochemical and physical specimen manipulation.

Purpose of the Study:

• To develop a computational solution for addressing map anisotropy and particle misalignment caused by the preferred orientation problem in cryo-EM.
• To introduce spIsoNet, an end-to-end self-supervised deep learning software for improving 3D reconstructions in cryo-EM.

Main Methods:

• Development of spIsoNet, a self-supervised deep learning algorithm.
• Utilizing preferred-orientation views to recover molecular information from under-sampled orientations.
• Application to various biological systems including ribosomes, β-galactosidases, and hemagglutinin trimers.

Main Results:

• spIsoNet significantly improves angular isotropy and particle alignment accuracy during 3D reconstruction.
• Near-isotropic reconstructions were generated from datasets with limited views.
• The software demonstrated effectiveness on diverse biological samples, including a previously challenging hemagglutinin trimer dataset.

Conclusions:

• spIsoNet provides a general computational solution to the preferred orientation problem in cryo-EM.
• The software enhances map isotropy and particle alignment without requiring additional specimen preparation.
• spIsoNet is applicable to both single-particle cryo-EM and subtomogram averaging of preferentially oriented molecules.