RelionEdit
Relion is a widely used open-source software package for cryo-electron microscopy (cryo-EM) data processing and three-dimensional structure determination. It implements a Bayesian approach to refinement using a regularized maximum-likelihood framework, which helps stabilize reconstructions from noisy particle images. Developed by a team at the MRC Laboratory of Molecular Biology led by Sjors Scheres, Relion has become a standard tool in the cryo-electron microscopy workflow and in molecular structure determination more broadly.
Relion’s influence extends beyond its algorithms. It provides an end-to-end pipeline that covers image preprocessing, particle picking, two-dimensional and three-dimensional classification, and high-resolution refinement. The software integrates with common ancillary tools for motion correction, contrast transfer function (CTF) estimation, and map validation, and it relies on a reproducible workflow built around its metadata-centric file format and job-based structure. This design emphasizes auditability and collaboration, as researchers can share parameter sets, pipelines, and refinements with colleagues in a way that few other packages have matched.
History
Relion emerged from the structural biology community’s efforts to better exploit cryo-EM data through principled statistical inference. Its development at the MRC Laboratory of Molecular Biology drew on Bayesian statistics and maximum-likelihood ideas to address the intrinsic noise and heterogeneity of cryo-EM images. Since its initial public releases, Relion has evolved through successive versions that expanded its automation, improved convergence properties, and broadened its applicability to a wider range of specimens, from small proteins to large macromolecular assemblies. The project has been sustained by an active user and developer community and has influenced the design of subsequent software in the field, including alternative approaches to particle classification and refinement in other toolchains such as cryoSPARC and EMAN2.
Technology and methods
Relion’s core is a probabilistic refinement engine that iteratively updates estimates of particle alignments, orientations, and the three-dimensional map. Its approach combines several key techniques:
Bayesian inference and regularized maximum-likelihood refinement, which balance fidelity to the data with prior information to avoid overfitting. See Bayesian statistics and Maximum likelihood for related concepts.
End-to-end processing steps that mirror the typical cryo-EM pipeline: motion correction, CTF estimation, particle picking, two-dimensional classification, three-dimensional classification, and high-resolution refinement. Elements of this pipeline are designed to be repeatable and auditable, with intermediate results stored in a consistent metadata framework.
Auto-refinement and classification capabilities that try to optimize particle orientations and class assignments in a unified framework, enabling more objective handling of heterogeneity in the data.
Advanced refinement features such as per-particle defocus estimation, beam-tilt correction, and particle polishing, which improve map quality by accounting for imaging imperfections. These steps are integrated into the processing stream and are compatible with the broader cryo-EM ecosystem, including tools for map sharpening and validation. See cryo-electron microscopy and Fourier shell correlation for related concepts.
Hardware acceleration, notably GPU-based computation, which has made large-scale refinements more practical for many laboratories. This has helped democratize access to high-end structure determination by leveraging common computing resources. See Graphics processing unit for background on the hardware aspect.
Output and validation metrics, including Fourier shell correlation (FSC) and related criteria, that provide objective measures of resolution and map quality. See Fourier shell correlation.
Relion’s design emphasizes compatibility with the broader ecosystem of cryo-EM software, as well as clear documentation and tutorials that help new users adopt best practices in refinement and validation. The software’s STAR-like metadata framework and its reproducible workflows are often cited as improvements for ensuring that published structures can be independently reproduced by other laboratories.
Impact on the field
Relion has helped standardize many aspects of the cryo-EM workflow, contributing to higher reproducibility and broader participation in structural biology. Its emphasis on rigorous statistical inference and automated workflows has reduced some of the arbitrariness in manual refinement, enabling labs to achieve high-resolution reconstructions with more consistent methodologies. The software’s influence is evident in the number of published structures that rely on Relion for processing workflows, and in how many laboratories have trained students and staff to implement a transparent, shareable pipeline for structure determination. See cryo-EM structure determination and 3D reconstruction for context on the broader field.
The Relion ecosystem also fostered collaboration among laboratories, software developers, and funding bodies, helping align computational infrastructure with the scientific questions being pursued. As cryo-EM projects scale in size and complexity, Relion’s emphasis on repeatability and validation has become an important benchmark for evaluating new methods and for teaching best practices in image analysis. See MRC Laboratory of Molecular Biology and Sjöres Scheres for historical and institutional context.
Controversies and debates
Like any influential software platform, Relion sits within a landscape of competing tools and methodological choices, and several debates have arisen within the community:
Open science, reproducibility, and access to resources. Proponents of open, community-driven software argue that Relion’s public availability and transparent workflows promote reproducibility and accelerate discovery. Critics sometimes raise concerns about the learning curve and required computational resources, which can favor well-funded laboratories and institutions with ready access to hardware and staff. Advocates for efficient use of public resources stress that robust, reproducible pipelines—whether implemented in Relion or alternative tools—maximize the return on investment in science. See open-source software and reproducibility in science.
Algorithmic priors, bias, and validation. Because Relion relies on priors and regularization to stabilize refinements in the presence of noisy data, there is ongoing discussion about how these choices influence the resulting maps. The community emphasizes independent validation, cross-method verification, and careful map versus model interpretation to guard against overfitting or bias introduced by priors. This debate intersects with broader questions about how best to balance data-driven inference with prior expectations. See Bayesian statistics and Fourier shell correlation.
Competition and ecosystem dynamics. Relion exists alongside other software packages that offer different strengths, such as alternative classification strategies or more automated pipelines. Debates often center on trade-offs between robustness, speed, and user control. The existence of multiple toolchains is generally viewed as a healthy ecosystem that spurs innovation, though some labs may prefer one framework over another for practical reasons. See cryoSPARC and EMAN2 as related options.
Diversity, inclusion, and opportunity in science. In wider science policy discussions, some critics argue that attention to diversity and inclusion diverts attention from merit-based funding and rapid progress. Proponents counter that diverse teams bring broader perspectives, reduce groupthink, and improve problem-solving, which can accelerate breakthroughs in difficult computational problems such as high-precision map interpretation. In technical communities, advocates for evidence-based policy emphasize that merit and opportunity can be advanced simultaneously through fair, transparent processes that reward capability and performance. See diversity in science and meritocracy.
Woke criticisms of scientific funding and culture are sometimes invoked in debates about Relion and cryo-EM workflows. From a practical standpoint, most practitioners judge that focusing on rigorous validation, reproducibility, and efficient use of resources serves science best, while treating merit and opportunity as the primary criteria for collaboration and advancement. Critics who dismiss these concerns as ideological caricatures often overlook the empirical benefits of objective standards, peer review, and verifiable results in keeping research on track and accountable.