BERIL Knowledge Wiki

Dataset

Protein Data Bank

Well corroborated. AI-generated wiki from AI-conducted research reports. Partially reviewed by scientists. See About This Wiki.

Sources cited
3 project reports and 1 cross-project digest, listed in the margin
Linked from
4 pages4 topics

Aliases: PDB. The Protein Data Bank supplies the experimental structures in the KBase Data Lakehouse inventory. The atlas counts ~253K PDB experimental structures alongside 241,070,489 AlphaFold predicted structures, 215,130,942 UniProt proteins, and 475,217,233 UniRef100, 188,848,220 UniRef90 and 60,315,044 UniRef50 clusters, with the UniRef counts dated 2026-01 1. Predicted AlphaFold models therefore far outnumber experimental PDB entries in the platform’s structural holdings 1.

Tables in the KBase Data Lakehouse

Two PDB collections are catalogued 1:

  • kescience_pdb: table pdb_entries, described as a kescience copy of the PDB experimental structures 1.
  • refdata_pdb: tables pdb_entries and pdb_uniprot_mapping, holding the PDB experimental structures plus a UniProt mapping 1.

Use in PaperBLAST

The PaperBLAST collection contains 12.4 million rows across 14 tables. It links genes to papers by text mining PubMed Central full-text articles, and it adds curated annotations from 13 databases and structural data from the PDB 2.

PaperBLAST holds site annotations for 132,179 PDB structures, with 2.1M site records across 4 types: binding (1.69M), functional (182K), modified (112K) and mutagenesis (104K) 2. The most frequent of its 48,991 unique ligands are zinc ions (65K sites), chlorophyll A (58K), calcium ions (53K) and heme (44K) 2.

Query pitfalls

The following documented caveats apply when querying the PDB tables 3:

  • r_work and r_free are NULL for non-X-ray methods (EM, NMR), and resolution is NULL for NMR. Apply filters by experimental method before averaging these fields 3.
  • pdb_uniprot_mapping.pdb_beg and pdb_end are stored as STRING and can literally contain the string "None" for unmapped regions. Coerce or filter these values before casting them to numbers 3.
  • pdb_entries.organism reports only the first polymer entity, so a multi-organism complex appears as a single-organism entry 3.
  • pdb_entries has one row per entry, not one row per chain. Use pdb_uniprot_mapping for chain-level analysis 3.

Prospective use

The AMR pangenome atlas names AlphaFold and PDB structures as resources for a future structural analysis. In that analysis, AMR proteins would be cross-referenced against these structures to identify novel resistance folds. The report lists this as a proposed next step, not a result, and no novel folds have been reported from it 4.

Sources cited

  1. BERDL Data Atlas — Inventory, Topic Map, and Cross-Reference Synergiesberdl_data_atlas · cited 5× · primary source
  2. PaperBLAST Data Explorer — Literature Coverage Bias in Protein Sequence Spacepaperblast_explorer · cited 3×
  3. KBase Data Lakehouse: Common Pitfalls & Gotchaspitfalls · cited 5× · cross-project digest
  4. Pan-Bacterial AMR Gene Landscapeamr_pangenome_atlas · cited 1×