FAIR Structural Data in the Age of AI: From Deposition to Discovery

This Research Topic is part of our FAIR series. For other Research Topics in the series, please see below:FAIR Metabolomics: Open Datasets, Shared Standards, and the Road to AI-Ready DataFAIR RNA Data: Open Resources for Transcriptomics, Structure, and Regulatory Network ResearchFAIR Genomics: Standards, Repositories, and Reproducibility in Chromatin and Genome ResearchFAIR Mitochondrial Data: Open Resources for Integrative Research in Bioenergetics and DiseaseFAIR Cell Biology Data: From Image Archives to Interoperable Signalling DatasetsMaking Glycan Data FAIR: Addressing the Interoperability Gap in GlycoscienceStructural biology has long been a leader in open data culture; the Protein Data Bank (PDB), Electron Microscopy Data Bank (EMDB), and Biological Magnetic Resonance Data Bank (BMRB) represent decades of community commitment to data sharing that has transformed our understanding of molecular structure and function. Yet the explosive growth of AI-driven structure prediction, cryo-electron microscopy (cryo-EM), and integrative modelling is generating new data types, scales, and reuse demands, including the use of structural data as training and benchmarking material for machine learning (ML), that existing FAIR infrastructure is only beginning to address.This Research Topic explores the frontier of FAIR data in structural biology, biophysics, and molecular evolution, including best practices for depositing AI-predicted structures and their experimental validation data, metadata standards for integrative and hybrid structural methods (PDB-Dev), and FAIR practices for cryo-EM datasets in EMPIAR and EMDB. We also welcome contributions addressing FAIR sharing of solution-state biophysical data from small-angle X-ray scattering (SAXS), isothermal titration calorimetry (ITC), and surface plasmon resonance (SPR) experiments, and NMR datasets in BMRB. The SASBDB (Small Angle Scattering Biological Data Bank) has established important infrastructure for SAXS data, but raw data deposition and reporting standards remain inconsistent and warrant community attention.As foundation models for structure prediction reshape the field, the provenance, quality annotation, and licensing of structural training data must themselves meet FAIR standards. We welcome perspectives on what ‘AI-ready’ structural data means and how the community should define it, encompassing not only format and completeness requirements, but also the principles governing responsible data sharing, cross-repository interoperability, and the ethical reuse of publicly deposited data for ML. How structural data repositories can evolve to serve both human researchers and AI-driven computational workflows simultaneously is a critical open question this Topic invites the community to address. FAIR² Data Articles presenting rich, annotated structural datasets with interactive visualization portals are an ideal submission type.KEY THEMES• FAIR deposition standards for AI-predicted structures and experimental validation data• Metadata standards for cryo-EM and cryo-electron tomography (cryo-ET): EMPIAR, EMDB, and map quality annotation• FAIR practices for NMR data: BMRB completeness and interoperability with PDB records• PDB-Dev and integrative hybrid method structure deposition standards• Time-resolved and dynamic structural data: formats, deposition, and metadata• Sequence-structure-function datasets for molecular evolution research• AI training and benchmarking datasets built on FAIR structural data• Community definitions of ‘AI-ready’ structural data and implementation frameworks• Data sharing policies, licensing, and governance frameworks for ML reuse of structural data• Cross-archive interoperability and data pipelines for ML workflows Topic editors: Yoshiki Yamaguchi Source: https://www.frontiersin.org/research-topics/83091/fair-structural-data-in-the-age-of-ai-from-deposition-to-discovery