EMPIAR — the Electron Microscopy Public Image Archive — is a public repository, run by EMBL-EBI, for the raw 2D image data that underlies cryo-electron microscopy (cryo-EM) structures: unprocessed micrographs, movie frames, particle stacks, and tomographic tilt series. It sits alongside two other, better-known structural biology archives, and researchers depositing or reusing cryo-EM data need to understand how the three fit together, because they store fundamentally different objects: raw images (EMPIAR), derived 3D density maps (EMDB), and final atomic models (PDB).
What EMPIAR archives, specifically
Cryo-EM structure determination starts with a physical sample imaged by an electron microscope, which produces thousands to millions of individual 2D images (micrographs, or movie frames from direct electron detectors) of frozen-hydrated particles or tomographic tilt series. Image-processing software then aligns, classifies, and averages that raw data to reconstruct a 3D density map. EMPIAR is the archive for the input to that process — the raw and minimally processed image data (micrographs, movie stacks, particle picks, tilt series) — not the reconstructed map itself. Beyond conventional single-particle cryo-EM, EMPIAR also accepts data from related volume electron microscopy (vEM) techniques and soft/hard X-ray tomography, reflecting its role as a general raw-image archive for 3D structural imaging rather than a cryo-EM-only resource.
EMPIAR entries are large by research-data standards — a single cryo-EM dataset commonly runs into the terabyte range, which is precisely why a dedicated raw-image archive with high-throughput deposition and retrieval infrastructure exists, rather than expecting researchers to host this volume of data themselves or bundle it as a journal supplementary file.
EMPIAR, EMDB, and the PDB: three archives, three stages
Structural biology has settled on a three-tier public archiving model that mirrors the cryo-EM workflow itself, and each tier is governed by a different (though related) set of infrastructure:
- EMPIAR — raw 2D image data: micrographs, movie frames, particle stacks, tilt series. Hosted and developed at EMBL-EBI.
- EMDB (Electron Microscopy Data Bank) — the derived 3D reconstructions (density maps) computed from EMPIAR-class raw data. EMDB was established in 2002, a decade before EMPIAR, specifically to archive cryo-EM 3D reconstructions; development of EMPIAR began in 2012 to close the gap left by EMDB not accepting the raw imaging data those reconstructions were built from.
- PDB (Protein Data Bank) — the final atomic coordinate models fitted into an EMDB map (for cryo-EM structures resolved to near-atomic resolution), archived through the worldwide Protein Data Bank (wwPDB) collaboration.
A single published cryo-EM structure can therefore have accessions in all three archives simultaneously: an EMPIAR entry for the raw images, an EMDB entry for the map reconstructed from them, and a PDB entry for the atomic model built into that map. EMPIAR enforces this chain explicitly at deposition time: with limited exceptions, every EMPIAR entry must be associated with at least one EMDB entry — the raw image data submitted has to be the data actually used to produce a specific, cited 3D reconstruction, not an orphaned raw dataset with no corresponding map. This is a meaningfully different provenance model from a general-purpose repository: EMPIAR is not a place to park arbitrary microscopy images, it is specifically the raw-data record behind a deposited reconstruction.
This three-way split matters for anyone planning a cryo-EM data management plan: the raw images, the map, and the model are not interchangeable outputs, they are three distinct deposition obligations with three distinct accession numbers, and reviewers or funders checking data availability may expect all three to be present and cross-referenced.
Accession numbers and how entries are cited
EMPIAR entries use the accession format EMPIAR-##### (for example, EMPIAR-10009), and depositors and downstream users should cite entries in exactly that form. Each EMPIAR entry’s metadata cross-references its associated EMDB accession(s), and by extension the PDB entry (or entries) built from that map, so a reader arriving at any one of the three accessions can trace back through the full raw-data-to-map-to-model chain for a given structure.
What research EMPIAR supports
EMPIAR’s primary user base is structural biologists working in single-particle cryo-EM, cryo-electron tomography, and related volume EM methods — determining the 3D structures of proteins, protein complexes, viruses, and organelles from frozen-hydrated specimens. Beyond the depositing lab itself, EMPIAR’s raw data is a resource for the wider community in several distinct ways: methods developers use publicly deposited raw datasets as benchmarks for new image-processing, particle-picking, and 3D-reconstruction algorithms; structural biologists reprocess existing raw data with improved software to obtain higher-resolution reconstructions than the original analysis achieved; and reviewers, journals, and funders increasingly treat a linked EMPIAR entry as part of verifying that a published cryo-EM structure is reproducible from its underlying evidence, not just from the reported map. Reflecting that role, journals with structural-biology data policies — for example, Nucleic Acids Research’s data deposition standards — recommend depositing raw cryo-EM image data to EMPIAR alongside the corresponding EMDB and PDB depositions.
Submitting to EMPIAR
Deposition is handled through EMBL-EBI’s own EMPIAR submission tooling (documented in EMPIAR’s deposition manual), separate from the wwPDB OneDep system typically used for the combined map-and-model deposition to EMDB and PDB. Practical points for anyone preparing a submission:
- Link to an EMDB entry first, or in parallel. Because an EMPIAR entry normally must be associated with an EMDB accession, depositors typically coordinate the two submissions rather than submitting raw images with no corresponding reconstruction on record.
- Expect large transfers. Cryo-EM raw datasets commonly run to hundreds of gigabytes and often several terabytes per dataset; EMPIAR’s submission and retrieval infrastructure is built around EMBL-EBI’s high-throughput data infrastructure for exactly this reason, and depositors should plan for the bandwidth and time this requires rather than treating it like a routine file upload.
- Provide the metadata that makes raw data reusable, not just storable — microscope and detector parameters, imaging conditions, and how the deposited images map to the associated EMDB reconstruction, so a third party can actually reprocess the dataset rather than merely download it.
- Check current policy before assuming deposition is optional. Requirements have moved toward stricter expectations for cryo-EM raw-data deposition as archive capacity and community standards evolve (EMPIAR has grown from a niche resource at its 2012 launch to several thousand entries and multiple petabytes of stored data); depositors should check EMPIAR’s current submission policy directly rather than relying on a prior paper’s description of what was required at the time it was written.
How this fits into a research data management plan
For any lab or core facility running cryo-EM as a service, EMPIAR deposition is a concrete example of a broader RDM principle: raw data and derived outputs have different reuse value and often belong in different repositories with different rules, even when they describe the same experiment. Treating the raw micrographs as “the data” and the final PDB structure as “the output” understates what EMPIAR is for — the raw images are themselves a citable, reusable research object, not disposable intermediate files, and planning for their deposition (storage, transfer, and metadata capture) belongs in a project’s data management plan from the start of a cryo-EM project, not as an afterthought at publication.
Frequently asked questions
Is EMPIAR the same thing as EMDB?
No. EMDB stores the derived 3D density map reconstructed from raw images; EMPIAR stores the raw 2D images (micrographs, movie frames, particle stacks, tilt series) that the reconstruction was built from. EMDB was established first, in 2002; EMPIAR was created a decade later, in 2012, specifically to archive the raw data that EMDB never captured.
Do I need to deposit in EMPIAR, EMDB, and the PDB separately?
For a typical near-atomic-resolution cryo-EM structure, yes — they are three separate archives capturing three separate stages of the workflow (raw images, map, atomic model), each with its own accession number, though EMPIAR entries are required to reference their associated EMDB entry so the two stay linked.
Who is EMPIAR intended for?
Primarily structural biologists depositing cryo-EM, cryo-electron tomography, and volume EM raw data, and secondarily the wider community reusing that raw data — methods developers benchmarking new reconstruction algorithms, and researchers reprocessing older datasets with improved software.
Is depositing raw cryo-EM images to EMPIAR mandatory?
Practice varies by journal and funder; several journals with structural-biology data policies now recommend or require it, and the expectation has been tightening over time. Depositors should confirm current requirements with the specific journal or funder, and check EMPIAR’s own submission policy rather than assuming past requirements still apply.
