FAIRification of structural biology experimental metadata

From its inception in 1971 as the first open-access digital data resource in biology, the Protein Data Bank (PDB), the global macromolecular structure archive, has grown to over 220,000 richly annotated 3D structures of proteins, DNA, RNA and their complexes. Founded in 2002, the Electron Microscopy Data Bank (EMDB) is a public repository of over 37,000 cryogenic-sample Electron Microscopy (cryoEM) volumes, often associated with 3D structures found in the PDB and representative tomograms of macromolecular complexes and subcellular structures. The Worldwide Protein Data Bank (wwPDB; wwpdb.org) partners,  including the founding member, the Protein Data Bank in Europe (PDBe) and the EMDB team, manage these core structural biology archives.
The free availability of 3D structure data has enormously impacted fundamental biology, biomedicine, biotechnology, and bioenergy by facilitating a mechanistic understanding of biological processes. Recent groundbreaking advances in structure prediction AI methods (e.g. AlphaFold) have demonstrated the importance of archiving and providing free access to data based on the Findable, Accessible, Interoperable and Reusable (FAIR) and Fairness, Accuracy, Confidentiality, Transparency (FACT) principles.
3D structures are mainly determined using macromolecular crystallography, nuclear magnetic resonance spectroscopy, single-particle cryoEM, and micro-crystal electron diffraction. Advances in time-resolved methods, X-ray free electron lasers, and new integrative methods are providing insights into ever-more complex biological systems at the atomic level, and cryogenic Electron Tomography is allowing studies of macromolecular machines “caught in the act” inside frozen cells.
All these datasets are deposited, validated, and curated using the wwPDB OneDep system developed by the wwPDB partnership. This system uses the International Union of Crystallography adopted extensible PDBx macromolecular Crystallographic Information Framework (PDBx/mmCIF) data dictionary and standard ontologies to support interoperability and reusability of structure data. The wwPDB partners, in collaboration with a group of community experts (wwPDB PDBx/mmCIF Working Group), coordinate PDBx/mmCIF dictionary development. The OneDep system has dramatically improved the coverage, integrity, and overall quality of PDB and EMDB data but has seen limited and incomplete deposition of experimental and analytical metadata accompanying the individual raw experimental datasets. The free availability of the experimental and analytical metadata via PDB and EMDB has the same potential to propel the development of advanced AI methods, driving more efficient experimental setups, reducing costs and accelerating experimental structure determination to facilitate life science research.
This project brings together all relevant stakeholders at the experimental facilities, structure determination software packages and wwPDB to address this shortcoming by developing an automatic data harvesting framework to transfer the experimental metadata through structure determination pipelines to the OneDep system. This new framework will be open access and can be adopted at other major facilities or laboratory-based instruments. The partner synchrotron and national cryo-EM facilities will be able to capture the metadata (such as beamline or microscope specifications and setup parameters, e.g. diffraction geometry and magnification used to collect data on the microscope) in a well-defined schema and transfer it alongside the datasets pre-processed through the structure determination software to the wwPDB OneDep system. The systematic archiving and free and standardised access to these datasets will accelerate the development of AI methods for better experimental setup and early evaluation of experiments to improve success rates, leading to efficient use of experimental facilities and accelerating experimental structural biology, facilitating basic and translational life science research at molecular details.