{
  "prepared_at_utc": "2026-09-09T05:06:25.614359+00:00",
  "selection": "Retrospective 20-gene metabolic-enriched expansion, excluding the first 41 FlyBase gene IDs. Includes GO/function, localization and name-only tiers.",
  "original_export": {
    "filename": "post-processed-2026_02_28k.xml",
    "sha256": "1268bb8b47e43df6940d4de5cd4839f77010a128d45354dba5177ffbe49ccc8e",
    "records": 28553,
    "matched_current_primary_accessions": 123,
    "additional_accessions_outside_published_list": 29,
    "additional_api_responses": 29,
    "primary_accession_join_only": true
  },
  "current_taxonomy_index": {
    "url": "https://rest.uniprot.org/uniprotkb/stream?query=taxonomy_id%3A7227&format=tsv&fields=accession%2Cid%2Corganism_id%2Cgene_names%2Clength%2Creviewed%2Cxref_flybase",
    "records": 42881,
    "retrieval_date_utc": "2026-09-09"
  },
  "reused_sources": {
    "projects/PROTNLM_EVALUATION/fly-benchmark/manifest.json": "0595d1562857a4d2fd2c9d9f69aaea48d96d3e5159faf632e5ad63be81290356",
    "projects/PROTNLM_EVALUATION/fly-benchmark/predictions.jsonl.gz": "a4406863de6d9021bde9721788fb8ebef5d12c4cdadc34617fc0d1889314a021",
    "projects/PROTNLM_EVALUATION/fly-benchmark/uniprot.jsonl.gz": "44a1ccfb6ca8639ed0785d61c9524f892f0c47efda87d748557c8d00144f2e1d",
    "projects/PROTNLM_EVALUATION/fly-benchmark/flybase-identifiers.tsv.gz": "d48298573536bbd049c082691f6631e732c18d31bc310abd777c0b6996518a17"
  },
  "checksums": {
    "original-fly-predictions.xml": "98008579294ff14d368c75ab339a3334cd4fad0d6c85d11703a2a02416460757",
    "additional-api-responses.jsonl.gz": "9565605927ad81af78a40b07e3628bbbfa46d20823c5fa1ca43408004fccbd62",
    "uniprot-taxonomy-index.tsv.gz": "38e77decf392c849206a109915cf6ca589beebe78ef8509b2a250e2026d3fa93"
  },
  "source_urls": {
    "prediction_api": "https://rest.uniprot.org/uniprotkb/protnlm/{accession}",
    "uniprot_api": "https://rest.uniprot.org/uniprotkb/{accession}.json"
  },
  "caveat": "Published-subset records reuse the 2026-09-08 snapshot; additional original-export records were fetched on 2026-09-09 UTC. XML sequence/taxon/date placeholders are not biological inputs. Current sequences do not establish prediction-time inputs. Existing annotations guide selection, not correctness."
}