{
  "schema": "massspec-science-example/v1",
  "split": "validation",
  "selection": "First available archived GLACIER validation query among early source records with precursor m/z below 260, at least eight peaks, [M+H]+ adduct, and at most 22 heavy atoms; selected before checking the target rank",
  "source": {
    "candidates_sha256": "9536a330cba24c399271364889119f90124579400dba14e9f78630c75c13c73a",
    "dataset_sha256": "50cfdd1d22f79543c59555f9ce43c6893bd788a19a42b21fb4e2e3a54673c06a",
    "repo_id": "roman-bushuiev/MassSpecGym",
    "revision": "d2e86d0c3bd905a6d578c0dd6053ed2bd41f9c2a"
  },
  "source_url": "https://huggingface.co/datasets/roman-bushuiev/MassSpecGym",
  "source_license": "MIT",
  "source_attribution": "MassSpecGym developers, 2024",
  "source_file_sha256_from_manifest": {
    "validation.jsonl": "26788b56a3cb9d19c265c925504a4ce4e77e4c6b65b2a11068be6f24d15bf70f",
    "validation_candidate_pools.jsonl": "7499330e87aedc2716c485c8fdf6205522ca7d0e7cff7f7ef2cd7d4fd87780e2"
  },
  "source_record_sha256": {
    "validation_query_line": "a27241c8f1ed40face95a42042b880a605f3742ce4e44da9f6a698a6e2758bfe",
    "validation_candidate_pool_line": "315b1281a556923d64c90c93c7c4296cc95ecd55021ad88e3c160ea14da95571",
    "validation_ranking_line": "27d93c458f3a0e5159b1b09df853f642b547c30b1f89a35909aef4ecef90fcd3"
  },
  "ranking_source": {
    "run": "glacier_full_validation/scored_entropy_all/rankings.jsonl",
    "sha256": "cb65ceead229f76d7a09ab5cd7f3d6e4d55b85e4da59c6c0b1e36d1fab8cbe97",
    "reference": "Official GLACIER MassSpecGym no-CF checkpoint; archived validation replay",
    "query_id": "MassSpecGymID0079471"
  },
  "query": {
    "adduct": "[M+H]+",
    "candidate_pool_id": "055da61389010f0cee75347294d779b03f74e95cf289551ba970ca20482b5cb8",
    "collision_energy": 20.0,
    "instrument_type": "QTOF",
    "intensities": [
      0.13013013013013014,
      0.011011011011011011,
      0.02002002002002002,
      0.01001001001001001,
      0.09109109109109109,
      0.05005005005005005,
      0.14214214214214213,
      0.8348348348348348,
      0.06906906906906907,
      0.03203203203203203,
      0.021021021021021023,
      0.016016016016016016,
      0.015015015015015015,
      0.16916916916916916,
      0.014014014014014014,
      0.016016016016016016,
      0.08108108108108109,
      0.043043043043043044,
      0.06906906906906907,
      0.23923923923923923,
      0.042042042042042045,
      1.0,
      0.02002002002002002,
      0.04504504504504504,
      0.012012012012012012,
      0.06706706706706707,
      0.01001001001001001,
      0.02002002002002002,
      0.013013013013013013,
      0.06306306306306306,
      0.05005005005005005,
      0.014014014014014014,
      0.036036036036036036,
      0.018018018018018018,
      0.2012012012012012,
      0.01001001001001001,
      0.06706706706706707,
      0.25325325325325326,
      0.05205205205205205,
      0.02002002002002002,
      0.07907907907907907,
      0.04804804804804805,
      0.014014014014014014,
      0.1981981981981982,
      0.1021021021021021
    ],
    "mzs": [
      41.026001,
      42.008673,
      42.021249,
      44.01309,
      52.018175,
      52.030752,
      53.026001,
      54.010016,
      54.033826,
      55.041651,
      64.018175,
      67.017841,
      68.024324,
      68.0369,
      68.997106,
      69.992355,
      78.010016,
      78.018569,
      79.017841,
      80.0369,
      81.020915,
      82.004931,
      82.039974,
      93.020915,
      94.013484,
      94.02874,
      95.008733,
      95.035223,
      95.047799,
      96.020581,
      96.031814,
      106.016164,
      107.023989,
      108.008005,
      108.031814,
      109.01583,
      111.042713,
      123.042713,
      134.011079,
      139.013818,
      151.037628,
      154.024717,
      169.035616,
      180.003982,
      197.030531
    ],
    "precursor_mz": 197.0305310981,
    "query_id": "MassSpecGymID0079471",
    "target_formula": "C6H4N4O4",
    "target_inchikey": "VTWQUFUBSCXPOW",
    "target_smiles": "Nc1nnc(-c2ccc([N+](=O)[O-])o2)o1"
  },
  "candidate_count": 256,
  "shown_candidates": [
    {
      "candidate_id": "8b11ff52e71cf5fa8993ca55afdb8608ae500a3bb7ffa53d73259dee39ee88a9",
      "smiles": "Nc1nnc(-c2ccc([N+](=O)[O-])o2)o1",
      "display_name": "A",
      "rank": 22
    },
    {
      "candidate_id": "74a2e39f2d9f28359b1ea0281a6ccd1c92a0acdf4c96081dfbc3468d8e4450ac",
      "smiles": "O=c1[nH]c(=O)n2c([N+](=O)[O-])ccc2[nH]1",
      "display_name": "B",
      "rank": 1
    },
    {
      "candidate_id": "99d5a97da81eb9878c1f3b1164bfd1b3b382c18ca1118496ffe5e90443edab62",
      "smiles": "O=c1[nH]c(=O)c2cc([N+](=O)[O-])cn2[nH]1",
      "display_name": "C",
      "rank": 2
    }
  ],
  "correct_candidate_id": "8b11ff52e71cf5fa8993ca55afdb8608ae500a3bb7ffa53d73259dee39ee88a9",
  "correct_candidate_rank": 22,
  "notes": [
    "Candidate structures are three examples from the actual mass-compatible pool.",
    "All plotted peaks and intensities are from the materialized validation record; intensities are normalized to the largest peak.",
    "The actual target is rank 22. This figure shows a retrieval error and does not claim benchmark-wide performance.",
    "No test sample, test label, or trained weight is included."
  ]
}