{
  "build_source_tree": {
    "scope": "fingerprinted database-build code, configuration, environment lock, and notices",
    "sha256": "e7dc0d2be039c556b9417fa4ffe04c1a5ab42c05e5c35f7e83d5443332a9cb9d"
  },
  "database_release": {
    "doi": "10.5281/zenodo.21443919",
    "state": "published",
    "url": "https://zenodo.org/records/21443919",
    "version": "1.0.2",
    "zenodo_concept_record_id": "21367798",
    "zenodo_record_id": 21443919
  },
  "inputs": [
    {
      "bytes": 140,
      "profile": "curated",
      "sha256": "8a706079542b2d50ec9d9040b26fce59a4152b0863a5e2ab86f0cad392c4d9cd",
      "table": "img_location"
    },
    {
      "bytes": 23557846,
      "profile": "curated",
      "sha256": "46f3bb8c5dd376ffa76d78717baff872198403b38082016620abae8612c75883",
      "table": "preferred_taxonomy"
    },
    {
      "bytes": 41707182,
      "profile": "curated",
      "sha256": "68cbdaef571f2828d837f617e7f51084b4812afbe610e6b27ac784a43837c855",
      "table": "sequences"
    },
    {
      "bytes": 71311364,
      "profile": "curated",
      "sha256": "61ab3ec8f75a7307af3c685d8a78c97e56feaf81d56b3e1de98be2a8110b5ea2",
      "table": "source_records"
    },
    {
      "bytes": 74358089,
      "profile": "curated",
      "sha256": "305b996f9258d6938dcc7717dc8de6edc6a6c2aa61de110e4a0e04b34470121c",
      "table": "taxonomy_assignments"
    },
    {
      "bytes": 177700,
      "profile": "img",
      "sha256": "8bde59cb14ad4d27efc5e8a8b6b898eb71fdf9983200b446bc0d654faed5597c",
      "table": "img_location"
    },
    {
      "bytes": 57470344,
      "profile": "img",
      "sha256": "4521d5ac0b53781f4d45783c2f3b2e2c3c11ac10e7e858132a8fdcae77e6ff9f",
      "table": "preferred_taxonomy"
    },
    {
      "bytes": 95384025,
      "profile": "img",
      "sha256": "204b4c516201cec56480aa61602a6e9e714ecf2ffb30187de264b6f188bbd7ce",
      "table": "sequences"
    },
    {
      "bytes": 166695466,
      "profile": "img",
      "sha256": "27d07b09d565c130d3209b81965e72cc812b400f8e948f8af4f53e75287ea80c",
      "table": "source_records"
    },
    {
      "bytes": 192133268,
      "profile": "img",
      "sha256": "e22e5c883ad1e1de0608ded05fa89dbf8db41937ea31b62638af863e40633d77",
      "table": "taxonomy_assignments"
    }
  ],
  "outputs": [
    {
      "bytes": 1238,
      "path": "docs/data/database_composition.tsv",
      "sha256": "c4843871f2f46ee2f924ac61ca8c0c40653ed78a6e2c0ce36ff58c934ef31465"
    },
    {
      "bytes": 1182,
      "path": "docs/data/database_sources.tsv",
      "sha256": "6b34395c748dd3a7401cbab47fadc454ba874ea3bf58a0fb8f322dd10ece051a"
    },
    {
      "bytes": 468,
      "path": "docs/data/database_taxonomy_resolution.tsv",
      "sha256": "696aa5778842b805bd26c80efe95ad65cd9533da5db8db9c7fc4ee1a303c95d6"
    },
    {
      "bytes": 11047,
      "path": "docs/data/database_taxonomy_lineages.tsv",
      "sha256": "c5da8e20671b756b6e87dd7e6c604f36751e76559ba77175b5b4020639dc9fc2"
    },
    {
      "bytes": 273684,
      "path": "docs/data/img_taxon_locations.parquet",
      "sha256": "635ab5ef68affe013ce1e98c616f94ce43a9d739f4a3303294c29ff2cf642263"
    },
    {
      "bytes": 119419,
      "path": "docs/data/img_location_sites.parquet",
      "sha256": "0ad9a008d658afb9f24b6768b03b9b09bdb017cc3b5d0ba57b709274f53e8dc1"
    }
  ],
  "privacy": {
    "excluded_source_metadata": "all non-allowlisted IMG metadata fields",
    "img_location_columns": [
      "taxon_oid",
      "latitude",
      "longitude"
    ]
  },
  "profiles": {
    "curated": {
      "archive": {
        "bytes": 362207547,
        "sha256": "5411b0f2c0d55f455c775cf350c4753b31e2b1e5af2752446edc59cef152a3b7",
        "url": "https://zenodo.org/records/21443919/files/ssuextract-db-curated-v1.0.2.tar.zst?download=1"
      },
      "counts": {
        "preferred_taxonomy": 609298,
        "sequences": 609298,
        "source_records": 683597,
        "taxonomy_assignments": 683597
      },
      "manifest_sha256": "13d399762ad7cd0b64486612cfc4e68b2ef1aa440ee55055eaf3c2ebead34b8f",
      "markers": {
        "16S": {
          "sequences": 416021,
          "source_records": 460239
        },
        "18S": {
          "sequences": 193282,
          "source_records": 223358
        }
      },
      "sources": [
        {
          "reference_source": "PR2",
          "source_records": 232042,
          "source_version": "5.1.1"
        },
        {
          "reference_source": "SILVA",
          "source_records": 451555,
          "source_version": "138.2"
        }
      ],
      "version": "1.0.2"
    },
    "img": {
      "archive": {
        "bytes": 881729306,
        "sha256": "24f0f9bad00273f8761ebb15f22931843939fa07d86f013ebba68eef557a2736",
        "url": "https://zenodo.org/records/21443919/files/ssuextract-db-img-v1.0.2.tar.zst?download=1"
      },
      "coordinate_coverage": {
        "located_source_records": 931810,
        "located_taxa": 30349,
        "location_table_rows": 32520,
        "marker_groups": [
          {
            "marker_group": "16S + 18S",
            "sites": 3445
          },
          {
            "marker_group": "16S only",
            "sites": 1336
          },
          {
            "marker_group": "18S only",
            "sites": 75
          }
        ],
        "sites": 4856,
        "source_records": 997180,
        "taxa": 32521,
        "taxa_without_location_row": 1
      },
      "counts": {
        "preferred_taxonomy": 1396949,
        "sequences": 1396949,
        "source_records": 1680777,
        "taxonomy_assignments": 1680777
      },
      "manifest_sha256": "6a246ef8f1f0dd71d385c56bcf57fc13907093ae3b67c80f1c84b45e36b9f24a",
      "markers": {
        "16S": {
          "sequences": 1067340,
          "source_records": 1299461
        },
        "18S": {
          "sequences": 329614,
          "source_records": 381316
        }
      },
      "sources": [
        {
          "reference_source": "IMG",
          "source_records": 997180,
          "source_version": "2025"
        },
        {
          "reference_source": "PR2",
          "source_records": 232042,
          "source_version": "5.1.1"
        },
        {
          "reference_source": "SILVA",
          "source_records": 451555,
          "source_version": "138.2"
        }
      ],
      "version": "1.0.2"
    }
  },
  "schema_version": 2,
  "taxonomy_policy": {
    "Archaea": "SILVA 138.2",
    "Bacteria": "SILVA 138.2",
    "Eukaryota": "PR2 5.1.1",
    "cross_domain_exact_sequence": "explicit ambiguity with native alternatives"
  },
  "taxonomy_summary": {
    "Archaea": {
      "target_rank": "phylum",
      "taxonomy_position": 2,
      "taxonomy_source": "SILVA",
      "taxonomy_source_version": "138.2"
    },
    "Bacteria": {
      "target_rank": "phylum",
      "taxonomy_position": 2,
      "taxonomy_source": "SILVA",
      "taxonomy_source_version": "138.2"
    },
    "Eukaryota": {
      "normalization": "PR2 compartment suffixes are removed from every taxonomy rank during import",
      "normalized_suffixes": [
        "apic",
        "chro",
        "chrom",
        "mito",
        "nucl",
        "plas"
      ],
      "target_rank": "supergroup",
      "taxonomy_position": 2,
      "taxonomy_source": "PR2",
      "taxonomy_source_version": "5.1.1"
    },
    "unresolved_definition": "no non-empty second taxonomy segment, or a second segment labeled unclassified or incertae sedis"
  }
}