2024BBSRC-NSF/BIO: Development of rich AI datasets for furthering life science research

The vision of this US RCSB Protein Data Bank/Protein Data Bank in Europe project is to automate data deposition and improve management of three-dimensional (3D) biostructure information stored in the Protein Data Bank (PDB) and the Electron Microscopy Data Bank (EMDB). 3D structures stored in the PDB archive provide a wealth of insights into biochemical and cellular function of biological macromolecules. Historically, each structure determination experiment resulted in a single 3D structure, which was deposited manually to the PDB and identified with a persistent identifier or PDB ID (e.g.,1vol). Primary experimental data underpinning PDB structures are stored in the PDB (macromolecular crystallography or MX: diffraction data) and EMDB (3D electron microscopy of 3DEM: electric Coulomb potential maps). Working with the four major MX and 3DEM structures determination software providers (Phenix, CCP4, CCP-EM, and Global Phasing), we will enable automated one-at-a-time deposition of individual MX and 3DEM structures directly from their programs, thereby avoiding laborious, error-prone manual data entry. We will also enable automated parallel deposition of multiple related MX and 3DEM structures, constituting an investigation. This work will benefit researchers and educators/students across the sciences by improving 3D biostructure data quality and completeness and grouping related structures, which together explain important biological phenomena.