RELION

Relion, short for Regularised Likelihood Optimisation, is an open‑source software package for single-particle analysis in cryo‑electron microscopy (cryo‑EM). It provides a Bayesian framework for reconstructing three‑dimensional structures from two‑dimensional projections, with emphasis on preventing overfitting through prior information and probabilistic modeling. It was developed by Sjors H. Scheres and colleagues at the MRC Laboratory of Molecular Biology and has become a widely used tool in the cryo‑EM community.

Key capabilities include automated particle picking, two‑dimensional classification, three‑dimensional classification, and high‑resolution three‑dimensional auto‑refinement. RELION also implements per‑particle and per‑image contrast transfer function refinement, beam‑induced motion correction through particle polishing, and post‑processing steps such as masking, map sharpening, and local resolution estimation. The software supports flexible workflows to handle both homogeneous samples and heterogeneous conformations.

RELION uses expectation–maximization algorithms to estimate the posterior distribution over particle orientations, class assignments, and structural parameters under Bayesian priors. It introduced gold‑standard refinement, which splits data into independent halves to assess resolution and reduce overfitting. The package supports multi‑class 3D classification to separate distinct structural states and is designed for parallel computation, including GPU acceleration, to handle large cryo‑EM datasets.

Impact and usage: RELION has become a standard in single‑particle cryo‑EM, contributing to many high‑resolution structures and shaping practices for validation and reporting. It coexists with other processing tools in the field and continues to be actively developed with new features and performance improvements.