RELION: Implementation of a Bayesian approach to cryo-EM structure determination¶
Why this mattered¶
RELION mattered because it helped turn single-particle cryo-EM from an expert-driven craft into a statistically principled, reproducible route to high-resolution structure determination. The paper’s central shift was to treat refinement as Bayesian inference over a noisy image-formation model, using maximum a posteriori optimization and explicit regularization rather than relying mainly on manually tuned alignment and classification parameters. This made refinement less dependent on user judgment, reduced overfitting, and gave practitioners a clearer statistical basis for interpreting particle orientations, reconstructions, and resolution estimates.
The “gold-standard” Fourier shell correlation procedure was especially consequential. By refining two independently split halves of the data and comparing them only at the end, RELION made it much harder for noise correlations and model bias to masquerade as structural detail. That practice became a field norm, helping establish more trustworthy resolution claims during the period when cryo-EM was moving toward near-atomic reconstructions.
After this paper, increasingly automated Bayesian refinement and classification made it practical to extract high-quality structures from heterogeneous, noisy particle populations at scale. RELION became one of the core software platforms of the cryo-EM “resolution revolution,” supporting later breakthroughs in membrane proteins, large macromolecular assemblies, and conformational-state analysis. Its importance was not just a better algorithm, but a change in epistemic standards: cryo-EM maps could be produced with less subjective tuning and evaluated with more rigorous safeguards against overinterpretation.
Abstract¶
RELION, for REgularized LIkelihood OptimizatioN, is an open-source computer program for the refinement of macromolecular structures by single-particle analysis of electron cryo-microscopy (cryo-EM) data. Whereas alternative approaches often rely on user expertise for the tuning of parameters, RELION uses a Bayesian approach to infer parameters of a statistical model from the data. This paper describes developments that reduce the computational costs of the underlying maximum a posteriori (MAP) algorithm, as well as statistical considerations that yield new insights into the accuracy with which the relative orientations of individual particles may be determined. A so-called gold-standard Fourier shell correlation (FSC) procedure to prevent overfitting is also described. The resulting implementation yields high-quality reconstructions and reliable resolution estimates with minimal user intervention and at acceptable computational costs.
Related¶
- cite → Maximum Likelihood from Incomplete Data Via the EM Algorithm — RELION adapts the EM algorithm's maximum-likelihood treatment of incomplete data to infer cryo-EM orientations, classes, and reconstructions from noisy particle images.
- enables ← Maximum Likelihood from Incomplete Data Via the EM Algorithm — The EM algorithm enables RELION through iterative maximum-likelihood estimation of hidden particle orientations and class assignments in cryo-EM.