Overview

Cryo-electron microscopy (cryo-EM) structure analysis encompasses the computational workflows that reconstruct three-dimensional macromolecular structures from tens of thousands of two-dimensional projection images acquired by an electron microscope. In single-particle cryo-EM, purified macromolecules embedded in vitreous ice are imaged in random orientations, and sophisticated algorithms determine the orientations, classify conformational states, and reconstruct a three-dimensional density map at near-atomic resolution. The resolution revolution of the past decade has made cryo-EM the method of choice for many large and flexible macromolecular complexes.

Methods

The computational pipeline begins with motion correction and contrast transfer function (CTF) estimation to correct for beam-induced movement and optical aberrations. Particle picking — increasingly performed by deep learning-based tools such as Topaz and crYOLO — identifies individual particle images. 2D classification groups particles into homogeneous classes, and 3D classification separates distinct conformational states. 3D reconstruction using Fourier-based methods (RELION, cryoSPARC) iteratively refines orientation parameters and builds the density map. Final resolution is assessed by the Fourier shell correlation (FSC) criterion.

Applications

Cryo-EM has resolved structures of the ribosome, spliceosome, membrane proteins, and viral capsids that were inaccessible to X-ray crystallography due to size or flexibility. The resulting density maps enable atomic model building, which is validated against known protein structure principles. Cryo-EM reveals the architecture of large cell structure and organelles, including the nuclear pore complex and mitochondrial respiratory supercomplexes. The technique relies on microscopy techniques for sample screening and has become integral to structural biology, with computational analysis as its core enabling technology.