{
  "files": {
    "json": {
      "bytes": 34892735,
      "path": "duf-protein-examples-2026-10-13.json",
      "sha256": "086fb9846de52d0d83d7efa5373883ebcc033425df93deb9f9c1ef7d9e841ea8"
    },
    "queries.jsonl": {
      "bytes": 19735393,
      "path": "duf-protein-examples-2026-10-13.queries.jsonl",
      "sha256": "29c55ae4774f9780ca50babae02100493f829c74be915ffd456d434323446e31"
    },
    "tsv": {
      "bytes": 422226,
      "path": "duf-protein-examples-2026-10-13.tsv",
      "sha256": "cf4ddd81c83233c2f29dff1f2bedb9dc5504ebb757daca4f1deb5aa433c7d1bd"
    }
  },
  "rows": {
    "by_status": {
      "HAS_SEQUENCE_EXAMPLE": 3106,
      "NO_SIBLING_CANDIDATE": 5190
    },
    "examples": 9378,
    "proteins": 8947,
    "total": 8296
  },
  "schema": {
    "tsv_fieldnames": [
      "pfam_id",
      "short_name",
      "status",
      "candidate_count",
      "example_count",
      "accessions",
      "unresolved_accessions"
    ]
  },
  "snapshot": {
    "date": "2026-10-13",
    "generated_at": "2026-10-13T00:47:41.727270+00:00",
    "id": "duf-protein-examples-2026-10-13",
    "input_snapshot_ids": {
      "worklist": "interpro-pfam-duf-2026-10-10"
    }
  },
  "source": {
    "acquisition_finished_at": "2026-10-13T00:46:57.923511+00:00",
    "acquisition_started_at": "2026-10-13T00:25:33.612774+00:00",
    "checkpoint_sha256": "29c55ae4774f9780ca50babae02100493f829c74be915ffd456d434323446e31",
    "cross_mech_sha256": "ff9ea3afb8fc6977a84f3e11aba487d622bd537390e6bd1ef8f12ce885b5b474",
    "cross_mech_snapshot": "cross-mech-duf-examples-2026-10-12",
    "license": "CC-BY-4.0 (UniProt sequences and metadata)",
    "license_url": "https://rest.uniprot.org/help/license",
    "limitations": [
      "Family membership is not experimental functional evidence.",
      "No domain coordinates are inferred from whole-protein sequences.",
      "No sibling candidate does not mean UniProt has no family members.",
      "Inactive, merged and isoform accessions are not silently substituted.",
      "Coverage is limited to the named source repositories and pinned commits."
    ],
    "mechs": {
      "AntibioticMech": {
        "commit": "18165f54cc13f9321330e34bdab71fd5bc03b05b",
        "data_dir": "data/antibiotics",
        "records_scanned": 2939,
        "repository": "https://github.com/CultureBotAI/AntibioticMech",
        "rows_before_uniprot": 0
      },
      "CellStructureMech": {
        "commit": "937ba6254eb987adcbd92e82b21b2ce8651dc5e3",
        "data_dir": "data/structures",
        "records_scanned": 909,
        "repository": "https://github.com/CultureBotAI/CellStructureMech",
        "rows_before_uniprot": 11
      },
      "CommunityMech": {
        "commit": "88050e999dd5699d08456f7f6495505618d6ead6",
        "data_dir": "kb/communities",
        "records_scanned": 458,
        "repository": "https://github.com/CultureBotAI/CommunityMech",
        "rows_before_uniprot": 0
      },
      "CultureMech": {
        "commit": "9976e05b88c89c6c6b5a4c1fd60ed0d6a86033a6",
        "data_dir": "data/normalized_yaml",
        "records_scanned": 15910,
        "repository": "https://github.com/CultureBotAI/CultureMech",
        "rows_before_uniprot": 0
      },
      "HabitatMech": {
        "commit": "679a57e04cd2581aec2cfbd1cdc5a78f09ec8a4a",
        "data_dir": "data/habitats",
        "records_scanned": 3208,
        "repository": "https://github.com/CultureBotAI/HabitatMech",
        "rows_before_uniprot": 0
      },
      "MediaIngredientMech": {
        "commit": "1742bfd450470f5479720ace96ad2851bf4cd9d0",
        "data_dir": "data/ingredients",
        "records_scanned": 2953,
        "repository": "https://github.com/CultureBotAI/MediaIngredientMech",
        "rows_before_uniprot": 0
      },
      "NaturalProductMech": {
        "commit": "e66f934111777c4ebfe69475d0520c0cfa074785",
        "data_dir": "data/natural_products",
        "records_scanned": 3115,
        "repository": "https://github.com/CultureBotAI/NaturalProductMech",
        "rows_before_uniprot": 0
      },
      "PathwayMech": {
        "commit": "e1c66917e753105c4bf5b7b026193fcefa2b000f",
        "data_dir": "data/pathways",
        "records_scanned": 233,
        "repository": "https://github.com/CultureBotAI/PathwayMech",
        "rows_before_uniprot": 0
      },
      "ProteinTraitsMech": {
        "commit": "70edd6f8458845f7d7783ac4f416df519f5e1de6",
        "data_dir": "data/traits",
        "records_scanned": 429293,
        "repository": "https://github.com/CultureBotAI/proteintraitsmech",
        "rows_before_uniprot": 37567
      },
      "TaxonMech": {
        "commit": "9e35388d8c663fa8b682551efd5c72cc218011bb",
        "data_dir": "data/taxa",
        "records_scanned": 625960,
        "repository": "https://github.com/CultureBotAI/TaxonMech",
        "rows_before_uniprot": 10
      },
      "TraitMech": {
        "commit": "62f45e1749fa3d3e5ca8b9a3a4ba5bb0361beb8b",
        "data_dir": "data/traits",
        "records_scanned": 1107,
        "repository": "https://github.com/CultureBotAI/TraitMech",
        "rows_before_uniprot": 33
      }
    },
    "name": "UniProt Consortium; sibling Mech accession references and PTM sequence registry",
    "ptm_direct_family_examples": {
      "commit": "70edd6f8458845f7d7783ac4f416df519f5e1de6",
      "examples": 7393,
      "qualification_statuses": {
        "LEGACY_OR_UNSPECIFIED": 7335,
        "QUALIFIED": 58
      },
      "records_inspected": 8296
    },
    "ptm_registry_discovery": {
      "family_protein_pairs": 1495,
      "mode": "all",
      "no_exact_current_entry": [],
      "proteins_checked": 12699,
      "without_target_family_count": 11283
    },
    "registry": {
      "commit": "70edd6f8458845f7d7783ac4f416df519f5e1de6",
      "path": "data/grounding/protein_registry.jsonl",
      "repository": "https://github.com/CultureBotAI/proteintraitsmech",
      "sha256": "1be0cf02b76b03a8fe9b80bcead9cd4a419157116eb8da038f266bcc33015eb9",
      "url": "https://github.com/CultureBotAI/proteintraitsmech/blob/70edd6f8458845f7d7783ac4f416df519f5e1de6/data/grounding/protein_registry.jsonl"
    },
    "uniprot_releases": [
      "2026_03"
    ],
    "url": "https://rest.uniprot.org/uniprotkb/search"
  }
}