Advancing FAIR and Sustainable 3DEM Data Ingestion, Validation, and Management at the Worldwide Protein Data Bank

Stephen K. Burley, M.D., D.Phil.�Director, RCSB Protein Data Bank�Rutgers, The State University of New Jersey�San Diego Supercomputer Center, University of California San Diego

Justin W. Flatt, Ph.D.�Biocurator, RCSB Protein Data Bank�Rutgers, The State University of New Jersey S²C² CryoEM Modeling & Structure Validation Workshop Oct. 30, 2025

​

​

rcsb.org

Outline

1. Frozen in Motion: 3DEM Then, Now, and Beyond – wwPDB milestones and evolution
2. Meeting the Data Deluge Challenge – Extended PDB IDs, PDBx/mmCIF, OneDep
3. Quality at Scale: Validation of 3DEM structures – Insights, Best practices, What’s Next

2

1. Frozen in Motion: 3DEM Then, Now, and Beyond – wwPDB milestones and evolution

3

Protein Data Bank (Established 1971)

• A single global archive for structural biology data
• Started with just 7 protein structures
• Now: ~245,000 structures freely available!
• Jointly managed by wwPDB partnership started in 2003
• Members: RCSB PDB (US), PDBe (Europe), PDBj (Japan), and PDBc (China); �plus EMDB (3DEM) and BMRB (NMR)

4

Protein Data Bank (1971) Nature New Biology 233, 223

Worldwide Protein Data Bank (2019) Nucleic Acids Research 47, D520–D528

X-ray structures led to the launch of archive

Cryo-EM not yet born when the PDB began

Annual PDB Structure Releases for 3DEM�Will Soon Eclipse Macromolecular Crystallography

5

Ratio (3DEM to MX released annually)

The 3DEM Data Resolution Revolution

Facilitated by

• Direct electron detectors
• Modern electron optics
• Improved data collection software
• Sophisticated image processing with improved algorithms
• FAIR and sustainable data management

6

300 keV • Direct Detector • RELION (1.9 Å)

Frozen In Motion: Then, Now, and Beyond

7

PDB 3DEM structures

EMDB Maps

AI-Fueled Future Needs Experts Now More than Ever

8

Cryo-EM Deposition Workshop 2004

Rutgers, USA

3DEM Validation Task Force 2010

Rutgers, USA

Cryo-EM Joint Map and Model Challenges Workshop 2017

SLAC, Stanford, USA

wwPDB: Data Stewardship by Domain Experts

• wwPDB ensures that data are freely and globally available
• Member sites include RCSB PDB, PDBe, PDBj, PDBc, EMDB and BMRB
• Sites collaborate on data deposition, biocuration, and validation via OneDep
• Each site provides different websites that offer complementary services and views of the data

​

9

• Previously intractable systems can often be studied at resolutions suitable for atomic model building
• AI/ML predictions can help (but be careful!)
• Experimental structures are essential for understanding biology

​

10

A New Era of Unprecedented Structural Complexity

Dendooven T. et al. Science Advances 9(30), eadg7480 (2023). doi:10.1126/sciadv.adg7480

Muir K. W. et al. Science 382(6675), 1184–1190 (2023). doi:10.1126/science.adj8736

​

A New Era of wwPDB Data Management

11

Vergara-Cruces Á. et al. Cell 187(5), 1145–1159.e21 (2024). doi:10.1016/j.cell.2024.01.036

1. Meeting the Data Challenge – Extended PDB IDs, PDBx/mmCIF, OneDep
2. Meeting the Data Deluge Challenge – Extended PDB IDs, PDBx/mmCIF, OneDep

12

​

​

​

​

​

​

​

Transitioning to Extended PDB IDs and PDBx/mmCIF Format to Support Growth

13

New PDB ID has 12-character length with “pdb_” prefix: pdb_[a-z0-9]{8}

Example: pdb_00006vxx

​

Available Transition Resources

14

Dedicated webpage describing the project

Documentation, FAQs, software tools, etc.

https://www.w wpdb.org/documentation/new-format-for-pdb-ids

Extended IDs in the public files – ready for testing

_database_2.database_id PDB

_database_2.database_code 4HHB

_database_2.pdbx_database_accession pdb_00004hhb

For the current entries, the extended PDB IDs are generated by adding prefix “pdb_0000” to their 4-character IDs:

All current archival files have been updated with the extended PDB ids, users can start testing /adopting the extended IDs now:

​

4HHB / pdb_00004hhb

Example files at GitHub – ready for testing

https://github.com/wwPDB/extended-wwPDB-identifier-examples

PDBx/mmCIF Data Format

15

PDBx/mmCIF: The master archival format of the PDB since 2014

Entries with extended PDB IDs will be available only in PDBx/mmCIF format

wwPDB is Evolving to Meet Community Needs

16

370+ IHMs already available at RCSB.ORG

Questions? Contact: brinda.vallat@rcsb.org

PDBx/mmCIF Data Resources

17

​

Documentation, FAQs, user guide available at https://mmcif.wwpdb.org/docs/user-guide/guide.html

• Software libraries, parsers, format converters, file editors
• Visualization tools: Chimera, Mol*, PyMol, Jmol, Coot, CCP4mg
• Refinement programs: Phenix, Refmac, Buster

​

Major structural biology software tools support PDBx/mmCIF

PDBx/mmCIF Powers Deeper Insight and Wider Impact for 3DEM

• No limits on atom count or polymer chains
• Continuously updated data dictionary
• wwPDB validation reports now available in PDBx/mmCIF format

Workflow:

EM extension dictionary:

18

18

Your data, fully realized:

Example:

_em_vitrification.instrument 'CRYOSOL VITROJET'

1. Meeting the Data Challenge – Extended PDB IDs, PDBx/mmCIF, OneD

Justin Flatt taking over (RCSB PDB Biocurator, 3DEM trained)

19

​

​

​

​

​

​

​

OneDep: wwPDB’s Unified Deposition System

20

Benefits

OneDep: A Global Tool for a Global Community

21

Benefits

22

Depositions Become “Easier” Inside OneDep

• Direct data harvesting from uploaded PDBx/mmCIF files
• Web UIs to collect metadata missing from uploaded files
• Issues PDB IDs for 3DEM models and EMDB IDs for 3DEM maps

​

​

Benefits

Tips for a Smoother Deposition Process

23

• Register for training events
• Prepare files and validate before deposition
• Ask questions anytime for guidance and support

General: deposit-help@mail.wwpdb.org

Specific: Log in to your session and select the Communication Page

• Provide accurate info for sequences and ligands
• Carefully review the validation report and address errors
• Share as you want others to share with you

​

​

​

Collaboration

24

Working Together For Accurate Records

SPT Labtech introduced the Chameleon in 2019

Request was not received to add the device to the pull-down until 2024

Please report any needed changes upon noticing!

vs.

As a result, existing entries need to be updated to specify chameleon

Collaboration

Complete Workflow Descriptions Enable Learning and Broader Adoption of More Robust Methods

25

25

Q96E09

PDB ID: 8so0/EMD-40644

Collaboration

26

Better Support for Composite Structure Depositions

Cumulative Growth

Each supporting map type (e.g., consensus EM volume, focused EM volume) must be clearly defined and linked to the composite model entry.

Challenges

Composite requirement: All supporting maps must be deposited separately!

Rise in composite model depositions

Majority: 3Å to 4Å (337 models)

Reuse of Common Metadata Streamlines Composite Depositions

27

Challenges

Use pdb_extract to Prepare Metadata Once for Reuse with Multiple Related Structure Submissions

28

Minimize effort, reduce human error

https://pdb-extract.wwpdb.org/

Solutions

Example: Streamlined Composite Structure Deposition

29

1. Fill out EM-specific metadata template
2. Upload model coordinates and metadata template
3. Enter full experimental sequences
4. Generate mmCIF for composite and map-only depositions

Upload to OneDep:

Solutions

RCSB PDB Support for 3DEM Structure Deposition

30

RCSB team handles all structures submitted from Americas and Oceania

RCSB PDB has processed over 16,219 EM entries to date — including 2858 for this year alone as of October 23, 2025. Always happy to help the 3DEM community!

Solutions

31

Evolving Validation Standards for 3DEM

32

Struct Dyn. 2020 Jan 24;7(1):014701. doi: 10.1063/1.5138589.

Never Take the FSC at Face Value

3. 2 Å

​

33

Always check local density and compare to similar-resolution maps

12 Å reported

Looking Good is not Always Being Right

34

Outliers can be refined away, creating deceptively perfect scores for imperfect structures

Looking “Too Good” is Easier to Spot

35

PDB ID: 8xbw/EMD-38231

Pintilie, G., Zhang, K., Su, Z. et al. Measurement of atom resolvability in cryo-EM maps with Q-scores. Nat Methods 17, 328–334 (2020). https://doi.org/10.1038/s41592-020-0731-1

Good Compared to What?

36

Modeling Ligands are Double the Trouble

37

Always verify that map fit, stereochemistry, and interactions are chemically sensible.

38

Let Q-scores be a Guide to Your Ligands

PDB ID: 6zcl/EMD-11166

Reprocessed and remodeled

Fit 1

vs.

2.8 Å

Q Struct: 0.62

Q Ligand: 0.52

Fit 2

Caution: ML Tools Can Distort Ligand Density!

39

Berkeley, R.F., Cook, B.D., Herzik Jr, M.A. (2024). Machine learning approaches to cryoEM density modification differentially affect biomacromolecule and ligand density quality. Front. Mol. Biosci, Sec. Structural Biology. Vol 11, https://doi.org/10.3389/fmolb.2024.1404885.

Extreme cases show ligands erased from the map

40

Bird’s Eye View of Ligand Quality Using Q-scores

Q = 0.52

Q = 0.20

41

Locking in Your Best Possible Model

• Validate early, validate often
• Leverage all available validation tools
• Model within the map’s resolution limits
• Visually inspect local fit across the structure
• Review official validation report and any correspondence carefully
• Note: Coordinates can be updated post-release, but the primary map cannot

The most valuable resource is the EM community

42

Outlook for Model Validation From Metrics to Machine Learning

43

Ongoing wwPDB Efforts to Support EM

RCSB.org Will Continue to Grow With the Community

• RCSB.org delivers
• >245,000 PDB structures
• >1 million Computed Structure Models (CSMs) from AlphaFold DB �and the ModelArchive
• RCSB.org data exploration and visualization tools used by many millions of researchers, educators, and students worldwide
• Provenance/reliability of both experimental structures and CSMs

44

Burley et al. (2023) Nucleic Acids Research 51, D488-D508

Recordings posted at PDB-101.rcsb.org

Follow Our Training Events and Grow With Us

US National Science Foundation (DBI-2321666), �National Institute of General Medical Sciences, �National Institute of Allergy and Infectious Disease, and �National Cancer Institute (NIH R01GM157729), and the US Department of Energy (DE-SC0019749)

Member of the �Worldwide Protein Data Bank �(wwPDB; wwpdb.org)

RCSB.ORG • info@rcsb.org

Core Operations Funding

Management

Follow us