{
  "bco_id": "http://data.glygen.org/GLY_000642",
  "checksum": "d56937dc640bd1c2c87fed8bfbfd158e",
  "bco_spec_version": "1.3.0",
  "provenance_domain": {
    "name": "Fruitfly Gene Names (NCBI RefSeq)",
    "version": "1.12.3",
    "created": "2022-08-30T15:40:57.216837",
    "modified": "2022-08-30T15:41:28.493870",
    "contributors": [
      {
        "orcid": "",
        "affiliation": "The George Washington University",
        "contribution": [
          "createdBy"
        ],
        "name": "Jeet Vora",
        "email": "jeetvora@gwu.edu"
      },
      {
        "orcid": "",
        "affiliation": "The George Washington University",
        "contribution": [
          "createdBy"
        ],
        "name": "Robel Kahsay",
        "email": "rykahsay@gwu.edu"
      },
      {
        "orcid": "",
        "affiliation": "The George Washington University",
        "contribution": [
          "createdBy"
        ],
        "name": "Karina Martinez",
        "email": "karinamartinez@gwu.edu"
      },
      {
        "orcid": "",
        "affiliation": "",
        "contribution": [
          "contributedBy"
        ],
        "name": "Not Applicable",
        "email": ""
      }
    ],
    "license": "https://creativecommons.org/licenses/by/4.0/"
  },
  "usability_domain": [
    "The Fruitfly Gene Names (NCBI RefSeq) dataset contains fruitfly [taxid:7227] gene Names/names (primary gene name and gene synonyms)  from NCBI RefSeq database.  If you use this dataset please provide proper attribution to NCBI RefSeq and GlyGen"
  ],
  "extension_domain": {
    "fhir_extension": [],
    "scm_extension": {
      "scm_repository": "",
      "scm_type": "git",
      "scm_commit": "",
      "scm_path": ""
    },
    "additional_license": {
      "data_license": "https://creativecommons.org/licenses/by/4.0/",
      "script_license": "https://www.gnu.org/licenses/gpl-3.0.en.html"
    },
    "dataset_categories": [
      {
        "category_value": "Drosophila melanogaster",
        "category_name": "species"
      },
      {
        "category_value": "Protein",
        "category_name": "molecule"
      },
      {
        "category_value": "Protein Information",
        "category_name": "tags"
      },
      {
        "category_value": "csv",
        "category_name": "file_type"
      },
      {
        "category_value": "Reviewed",
        "category_name": "status"
      },
      {
        "category_value": "internal",
        "category_name": "scope"
      }
    ]
  },
  "description_domain": {
    "keywords": [
      "protein gene names",
      "NCBI RefSeq"
    ],
    "pipeline_steps": [
      {
        "step_number": 1,
        "name": "make-protein-dataset.py",
        "description": "The script parses the NCBI RefSeq gpff files to extract the gene Names/names and then maps the NCBI RefSeq accesion to the UniProtKB Canonical accessions",
        "version": "",
        "prerequisite": [],
        "input_list": [
          {
            "access_time": "",
            "uri": "https://data.glygen.org/ln2data/releases/data/v-1.12.3/reviewed/fruitfly_protein_xref_refseq.csv",
            "filename": "fruitfly_protein_xref_refseq.csv"
          },
          {
            "access_time": "",
            "uri": "https://data.glygen.org/ln2data/releases/data/v-1.12.3/source/refseq_protein_all_7227.gpff",
            "filename": "refseq_protein_all_7227.gpff"
          }
        ],
        "output_list": [
          {
            "access_time": "",
            "uri": "https://data.glygen.org/ln2data/releases/data/v-1.12.3/reviewed/fruitfly_protein_genenames_refseq.csv",
            "filename": "fruitfly_protein_genenames_refseq.csv"
          }
        ]
      }
    ]
  },
  "execution_domain": {
    "script": [
      {
        "uri": {
          "access_time": "",
          "sha1_chksum": "",
          "uri": "https://github.com/GW-HIVE/glygen-backend-integration/blob/master/pipeline/integrator/make-protein-dataset.py",
          "filename": "make-protein-dataset.py"
        }
      }
    ],
    "script_driver": "manual",
    "software_prerequisites": [
      {
        "name": "Python",
        "version": "2.7.5",
        "uri": {
          "access_time": "2019-01-07T17:43:17-0500",
          "sha1_chksum": "",
          "uri": "https://www.python.org/download/releases/2.7.5/"
        }
      }
    ],
    "external_data_endpoints": [],
    "environment_variables": {}
  },
  "parametric_domain": [
    {
      "param": "",
      "value": "",
      "step": ""
    }
  ],
  "io_domain": {
    "input_subdomain": [
      {
        "uri": {
          "access_time": "",
          "sha1_chksum": "",
          "uri": "https://data.glygen.org/ln2data/releases/data/v-1.12.3/source/refseq_protein_all_7227.gpff",
          "filename": "refseq_protein_all_7227.gpff"
        }
      },
      {
        "uri": {
          "access_time": "",
          "sha1_chksum": "",
          "uri": "https://data.glygen.org/ln2data/releases/data/v-1.12.3/reviewed/fruitfly_protein_xref_refseq.csv",
          "filename": "fruitfly_protein_xref_refseq.csv"
        }
      }
    ],
    "output_subdomain": [
      {
        "mediatype": "csv",
        "uri": {
          "access_time": "",
          "sha1_chksum": "",
          "uri": "https://data.glygen.org/ln2data/releases/data/v-1.12.3/reviewed/fruitfly_protein_genenames_refseq.csv",
          "filename": "fruitfly_protein_genenames_refseq.csv"
        }
      },
      {
        "mediatype": "csv",
        "uri": {
          "access_time": "",
          "sha1_chksum": "",
          "uri": "https://data.glygen.org/ln2data/releases/data/v-1.12.3/reviewed/fruitfly_protein_genenames_refseq.csv.stat.csv",
          "filename": "fruitfly_protein_genenames_refseq.csv.stat.csv"
        }
      }
    ]
  },
  "error_domain": {
    "empirical_error": {},
    "algorithmic_error": {},
    "emperical_error": {}
  }
}