TSV · AUDITHigh quality
OpenPLBind Refined Set v2
Exact Kd/Ki complexes passing crystallographic, ligand, affinity-range and structure-quality rules.
- Records
- 727
- Base pass
- 1,242
Open release · 2021+ v2
Explore traceable structure–binding data.
OpenPLBind links experimentally resolved PDB complexes—including small-molecule and polymeric ligands—to literature-grounded Kd, Ki, and IC50 measurements, with auditable provenance from paper to structure.
Popular:
4,039Processed structures
4,039Activity records
402Refined Set records
1,939Source publications
2,822Ligand components
New
Refined Set v2 is available727 high-quality Kd/Ki complexes pass the combined experimental, ligand and structure-quality rules.
Explore Refined Set →Start exploring
Built around open scientific data.
Search the catalog, download familiar index formats, or review how every structure–activity link is validated.
Find data
Search activity records and filter by year, metric and export class.
Explore catalog →Get release files
Download PDBbind-style indexes, provenance tables and validation notes.
Open downloads →Understand the data
Read the identity-gated curation and coordinate preparation policies.
Read methodology →Featured datasets
Release highlights
TSVValidated
Primary activity index
One conservative scalar Kd, Ki or IC50 label per PDB entry, with source DOI and selection provenance.
- Records
- 4,039
- Coverage
- 2021–2026
PDB · CIF · SDFValidated
Classic activity subset
Scalar activity labels backed by classic single-component, non-covalent ligand structure exports.
- Records
- 3,024
- Class
- Classic
PDB · CIFPreserved
Extended activity subset
Scalar labels for polymer, branched and format-limited ligand cases preserved without unreliable small-molecule representations.
- Records
- 1,015
- Class
- Extended
Browse collections
Explore by activity type.
Jump directly into evidence-grounded subsets of the release.
Data catalog
Search the 2021+ release.
Find structures by PDB ID, ligand, DOI, activity type, release year or export class. Every record links back to its source publication.
| PDB ID | Activity | Ligand | Year | Resolution | Method | Class | Details |
|---|---|---|---|---|---|---|---|
| Loading release index… | |||||||
—
Release files
Download OpenPLBind v2.
Use corrected PDBbind-style labels or the complete machine-readable manifest in model-development workflows.
OpenPLBind_refined_set_v2.tsvFinal 727-entry Refined Set with selection fields86.2 KB ↓
OpenPLBind_refined_selection_audit_v2.zipFinal set, 1,242 base candidates, complete audit, rules and summary232.1 KB ↓
INDEX_openplbind_PL_data.4039Corrected compact labels for all 4,039 PDB entries427.6 KB ↓
training_manifest.tsvComplete machine-readable labels, assay context, provenance and paths2.87 MB ↓
structure_index.tsvPer-entry index of the publishable structure files565.8 KB ↓
openplbind_v2_release_notes.zipREADME, corrected-normalization summary and Refined Set summary3.7 KB ↓
Identity-gated curation
More than an affinity number.
OpenPLBind treats the evidence-supported correspondence between a crystallographic complex and an experimental assay as the annotation unit.
Refined Set v2 selection
A reproducible path from activity evidence to 727 high-quality complexes.
4,050primary activity rows
→1,242pass base rules
→727pass quality audit
- 01
X-ray only; resolution ≤ 2.5 Å and R-factor < 0.25.
- 02
Noncovalent nonpolymer ligand; allowed elements only and molecular weight < 1000 Da.
- 03
Protein-buried ligand fraction ≥ 0.15.
- 04
Exact Kd or Ki between 1 pM and 10 mM.
- 05
No detected completeness, nonstandard-residue, steric-clash or multiple-site violation.
Quality checks use fail-evidence semantics: only a detected violation excludes a base candidate. Rule-level failure counts may overlap.
- 01
Collect
Start from post-2020 PDB entries, normalize citations and retrieve source documents.
- 02
Frame
Resolve protein construct, mutation, ligand instance and complex identity.
- 03
Annotate
Extract assay-aware measurements with page-level evidence.
- 04
Validate
Apply identity, chemistry, coordinate and release consistency checks.
Scope note
“PDBbind-style” describes the familiar per-entry file layout and task semantics, not the original PDBbind biological-unit preparation. OpenPLBind retains all non-focal polypeptide chains from the first deposited model; it does not apply assembly expansion, structure repair or protonation, and it removes water and other nonpolymers.