DUFMech

Sources and schema

Complete family index, metadata and provenance (JSON)

Frozen inputs

Protein domain match dataset

pfam-uniprot-uniref90-2026-10-07: 2 frozen member rows across 1 families. This subset is not a complete family membership census.

Generated: 2026-10-07T07:39:51.164664Z. Worklist seed: interpro-pfam-duf-2026-10-05. SHA-256: 8915922e639601527d923bcba45420386add54180525fdb242f77e961437dcda.

Download frozen member rows (JSON) | Download checksum manifest (JSON)

Download member table (TSV)

Source: InterPro Pfam members, UniProtKB and UniRef. Pfam membership and match coordinates: InterPro, CC0. Protein metadata and UniRef mappings: the UniProt Consortium, CC BY 4.0. The figures render the recorded coordinates without inferring structure or function.

Protein sequence example dataset

duf-protein-examples-2026-10-13: 8,947 proteins; 3,106 families with sequences out of 8,296 audited families.

Frozen examples (JSON) | Manifest

UniProt Consortium, CC BY 4.0. Sequence examples are family-membership references, not functional claims.

Site build f6a3d5197ea761bc. Source links are pinned to the supplied or verified source commits. Input snapshot identifiers are listed in the footer. No live scientific data is fetched by these pages.

Retained reviews

Record schema

LinkML family schema documentation

Status and evidence

Seed status
A name-based classification of frozen InterPro/Pfam metadata. UNKNOWN CANDIDATE means the metadata explicitly says the function is unknown. KNOWN HISTORICAL DUF means a DUF name matched without that wording; this heuristic is not proof of a known function and does not establish experimental characterization. EX DUF means Pfam renamed the family away from a DUF/UPF name, as recorded in Pfam's previous identifiers; a rename usually follows published work on some members but is naming history, not proof of function for every member. Some EX DUF families' current names or descriptions still say the function is unknown; their classification basis lists this, and scoring does not treat them as characterized. FALSE POSITIVE TEXT HIT means the naming rules did not match.
Characterization
A separate evidence-scoring result. UNSCORED means no score has been calculated; it is not evidence that the family lacks a known function. Missing scores are not negative evidence. PARTIALLY CHARACTERIZED has partial support; KNOWN HISTORICAL DUF can inherit the seed label or reflect a known-function signal. Neither label implies experimental validation.
Evidence counts
Source-specific signals, not unique proteins, publications, or confidence scores. Known counts Rhea reactions and experimentally supported GO molecular-function annotations. Partial counts functional/domain assignments. Context counts structural, environmental, orthology, interaction and neighborhood signals; context alone does not assign function. Repeated annotations may not be independent.
Curation status
Reported separately from seed classification and characterization. SNAPSHOT ONLY means no curated metadata was supplied for this build. SEEDED denotes a source-derived record, not an expert-reviewed functional claim.
Missing counts
Not available means the source did not supply that counter. A zero is a reported count. Not scored means evidence counts are unavailable.