PLINDER documentation#

plinder

PLINDER is the Protein & Ligand INteraction Dataset and Evaluation Resource: a comprehensive, annotated, high-quality resource for training and evaluating protein-ligand and protein-protein structure models.

  • 611,930 protein–ligand systems, 803,190 proper ligands, and 3,811,623 protein–protein interfaces

  • Ligand and protein-interface annotations plus compact entry, chain, and representative tables

  • Automated curation pipeline to keep up with the PDB

  • Reusable ligand, pocket, and protein-interface similarities and representative assignments

  • Ranked deposited apo protein chains linked to holo ligands and protein-interface sides

  • Python APIs for release queries, coordinate reconstruction, and custom scoring

Getting started

Install PLINDER, select a release, and continue to the runnable guides.

Getting started
Data access

Query columns across the ligand, interface, entry, chain, and representative tables.

Query the release tables
Structures

Work with ligand systems, linked apo chains, protein interfaces, atom masks, and self-contained mmCIF files.

Access system files with PlinderSystem
Similarity and scoring

Use pairwise scores and representative covers, or compare custom mmCIF and FASTA inputs with PLINDER.

Guides
Data reference

Inspect release files, table relationships, column definitions, and coordinate provenance.

Dataset