Metadata¶

pandora.metadata extracts source-backed metadata (never derived — everything here comes straight from an mmCIF category) from a parsed Structure. See Functions for full signatures, and Schemas for every record's fields.

There's no pandora metadata CLI subcommand — collect_metadata() only runs internally, inside pandora curate. Everything below is library-only.

Collect everything at once¶

collect_metadata() is the orchestrator: it calls every extract_*() function below and bundles the results into one MetadataRecord.

from pandora.parsing import mmcif_to_structure
from pandora.metadata import collect_metadata

structure, _, _ = mmcif_to_structure("datasets/dev/mmcif/104m.cif")
metadata = collect_metadata(structure)

print(metadata.entry.title)
# SPERM WHALE MYOGLOBIN N-BUTYL ISOCYANIDE AT PH 7.0
print([ligand.comp_id for ligand in metadata.ligands])
# ['SO4', 'HEM', 'NBN']
print(len(metadata.taxonomies))
# 1

Extract one category at a time¶

Each extract_*() function reads one or a few specific mmCIF categories and is independently callable — reach for these when you only need one piece, not the full MetadataRecord.

from pandora.metadata.mmcif import (
    extract_entry_metadata,
    extract_entity_metadata,
    extract_ligand_metadata,
    extract_quality,
    extract_taxonomies,
    extract_taxonomy,
    extract_uniprot_mappings,
)

entry = extract_entry_metadata(structure)
print(entry.title, entry.doi)
# SPERM WHALE MYOGLOBIN N-BUTYL ISOCYANIDE AT PH 7.0 None

entities = extract_entity_metadata(structure)
print([(e.entity_id, e.entity_type) for e in entities])
# [('1', 'polymer'), ('2', 'non-polymer'), ('3', 'non-polymer'), ('4', 'non-polymer'), ('5', 'water')]

ligands = extract_ligand_metadata(structure)
print([(l.comp_id, l.name) for l in ligands])
# [('SO4', 'SULFATE ION'), ('HEM', 'PROTOPORPHYRIN IX CONTAINING FE'), ...]

quality = extract_quality(structure)
print(quality.experimental_method, quality.resolution)
# X-ray diffraction 1.71

taxonomies = extract_taxonomies(structure)
print([t.organism_scientific for t in taxonomies])
# ['Physeter catodon']

taxon = extract_taxonomy(structure)  # first taxonomy record, or None
print(taxon.organism_scientific)
# Physeter catodon

mappings = extract_uniprot_mappings(structure)
print([m.accession for m in mappings])
# ['P02185', 'P02185']

Read a raw category directly¶

Not every mmCIF category has a typed extractor yet. extract_metadata_category() is the escape hatch: it reads any category straight out of Structure.raw.

from pandora.metadata.mmcif import extract_metadata_category

rows = extract_metadata_category(structure, "_exptl")
print(rows)
# [{'entry_id': '104M', 'method': 'X-ray diffraction', 'crystals_number': '1'}]

Pass columns=[...] to restrict which fields come back. extract_metadata() is an older-name alias for the same function, kept for compatibility.