{
  "bco_id": "http://data.glygen.org/GLY_000628",
  "checksum": "4c1e57d6e541f5e8a1a6521445502cb2",
  "bco_spec_version": "1.3",
  "provenance_domain": {
    "name": "SARS-CoV1 Glycosylation Sites (UniCarbKB)",
    "version": "1.12.3",
    "created": "2022-03-31T15:05:49.377202",
    "modified": "2022-09-06T17:51:39.310277",
    "contributors": [
      {
        "orcid": "",
        "affiliation": "The George Washington University",
        "contribution": [
          "createdBy"
        ],
        "name": "Karina Martinez",
        "email": "karinamartinez@gwu.edu"
      },
      {
        "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@email.gwu.edu"
      },
      {
        "orcid": "",
        "affiliation": "Griffith University",
        "contribution": [
          "contributedBy"
        ],
        "name": "Matthew Campbell",
        "email": "m.campbell2@griffith.edu.au"
      }
    ],
    "license": "https://creativecommons.org/licenses/by/4.0/"
  },
  "usability_domain": [
    "This is a SARS-CoV1 Glycosylation Sites (UniCarbKB) dataset. It contains a list of SARS coronavirus (SARS-CoV-1) [taxonomy:694009] proteins with information on glycosylation sites and associated glycans (if available) from UniCarbKB. The listed proteins (UniProtKB accession) are part of the GlyGen UniProtKB SARS CoV1 Proteome Masterlist (https://data.glygen.org/GLY_000467). If you use this dataset, please provide proper attribution to UniCarbKB, EMBL-EBI-UniProtKB, and GlyGen. Cite: PMID:24234447, PMID:33237286, and PMID:31616925. This dataset can be used for obtaining a list of glycosylation sites for canonical glycoproteins from the UnicarbKB database."
  ],
  "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": "Reviewed",
        "category_name": "status"
      },
      {
        "category_value": "Proteoform",
        "category_name": "molecule"
      },
      {
        "category_value": "SARS coronavirus (SARS-CoV-1)",
        "category_name": "species"
      },
      {
        "category_value": "UniCarbKB glycosylation sites",
        "category_name": "tags"
      },
      {
        "category_value": "csv",
        "category_name": "file_type"
      },
      {
        "category_value": "internal",
        "category_name": "scope"
      }
    ]
  },
  "description_domain": {
    "keywords": [
      "proteins",
      "canonical",
      "glycosylation sites",
      "glycan",
      "UniCarbKB",
      "GlyTouCan"
    ],
    "pipeline_steps": [
      {
        "step_number": 1,
        "name": "make-proteoform-dataset.py",
        "description": "",
        "version": "",
        "prerequisite": [],
        "input_list": [
          {
            "access_time": "",
            "sha1_chksum": "",
            "uri": "http://data.glygen.org/ln2downloads/unicarbkb/01_11_2022/known_33064451_glygen.csv",
            "filename": "known_33064451_glygen.csv"
          },
          {
            "access_time": "",
            "sha1_chksum": "",
            "uri": "https://data.glygen.org/ln2data/releases/data/v-1.12.3/reviewed/sarscov1_protein_masterlist.csv",
            "filename": "sarscov1_protein_masterlist.csv"
          },
          {
            "access_time": "",
            "sha1_chksum": "",
            "uri": "https://data.glygen.org/ln2data/releases/data/v-1.12.3/source/classification.tsv",
            "filename": "classification.tsv"
          },
          {
            "access_time": "",
            "sha1_chksum": "",
            "uri": "https://data.glygen.org/ln2data/releases/data/v-1.12.3/source/glycan_properties.tsv",
            "filename": "glycan_properties.tsv"
          },
          {
            "access_time": "",
            "sha1_chksum": "",
            "uri": "https://data.glygen.org/ln2data/releases/data/v-1.12.3/reviewed/sarscov1_protein_allsequences.fasta",
            "filename": "sarscov1_protein_allsequences.fasta"
          },
          {
            "access_time": "",
            "sha1_chksum": "",
            "uri": "https://data.glygen.org/ln2releases/v-1.12.3/misc/aadict.csv",
            "filename": "aadict.csv"
          },
          {
            "access_time": "",
            "sha1_chksum": "",
            "uri": "https://data.glygen.org/ln2data/releases/data/v-1.12.3/reviewed/glycan_xref_unicarbkb.csv",
            "filename": "glycan_xref_unicarbkb.csv"
          }
        ],
        "output_list": [
          {
            "access_time": "",
            "sha1_chksum": "",
            "uri": "https://data.glygen.org/ln2data/releases/data/v-1.12.3/reviewed/sarscov1_proteoform_glycosylation_sites_unicarbkb.csv",
            "filename": "sarscov1_proteoform_glycosylation_sites_unicarbkb.csv"
          }
        ]
      }
    ]
  },
  "execution_domain": {
    "script": [
      {
        "uri": {
          "access_time": "",
          "sha1_chksum": "",
          "uri": "https://github.com/glygener/glygen-backend-integration/blob/master/pipeline/integrator/make-proteoform-dataset.py",
          "filename": "make-proteoform-dataset.py"
        }
      }
    ],
    "script_driver": "",
    "software_prerequisites": [
      {
        "name": "Python",
        "version": "2.7.5",
        "uri": {
          "access_time": "",
          "sha1_chksum": "",
          "uri": "https://www.python.org/download/releases/2.7.5/",
          "filename": ""
        }
      }
    ],
    "external_data_endpoints": [],
    "environment_variables": {}
  },
  "parametric_domain": [
    {
      "param": "",
      "value": "",
      "step": ""
    }
  ],
  "io_domain": {
    "input_subdomain": [
      {
        "uri": {
          "access_time": "",
          "sha1_chksum": "",
          "uri": "http://data.glygen.org/ln2downloads/unicarbkb/01_11_2022/known_33064451_glygen.csv",
          "filename": "known_33064451_glygen.csv"
        }
      },
      {
        "uri": {
          "access_time": "",
          "sha1_chksum": "",
          "uri": "https://data.glygen.org/ln2data/releases/data/v-1.12.3/reviewed/sarscov1_protein_masterlist.csv",
          "filename": "sarscov1_protein_masterlist.csv"
        }
      },
      {
        "uri": {
          "access_time": "",
          "sha1_chksum": "",
          "uri": "https://data.glygen.org/ln2data/releases/data/v-1.12.3/source/classification.tsv",
          "filename": "classification.tsv"
        }
      },
      {
        "uri": {
          "access_time": "",
          "sha1_chksum": "",
          "uri": "https://data.glygen.org/ln2data/releases/data/v-1.12.3/source/glycan_properties.tsv",
          "filename": "glycan_properties.tsv"
        }
      },
      {
        "uri": {
          "access_time": "",
          "sha1_chksum": "",
          "uri": "https://data.glygen.org/ln2data/releases/data/v-1.12.3/reviewed/sarscov1_protein_allsequences.fasta",
          "filename": "sarscov1_protein_allsequences.fasta"
        }
      },
      {
        "uri": {
          "access_time": "",
          "sha1_chksum": "",
          "uri": "https://data.glygen.org/ln2releases/v-1.12.3/misc/aadict.csv",
          "filename": "aadict.csv"
        }
      },
      {
        "uri": {
          "access_time": "",
          "sha1_chksum": "",
          "uri": "https://data.glygen.org/ln2data/releases/data/v-1.12.3/reviewed/glycan_xref_unicarbkb.csv",
          "filename": "glycan_xref_unicarbkb.csv"
        }
      }
    ],
    "output_subdomain": [
      {
        "mediatype": "csv",
        "uri": {
          "access_time": "",
          "sha1_chksum": "",
          "uri": "https://data.glygen.org/ln2data/releases/data/v-1.12.3/reviewed/sarscov1_proteoform_glycosylation_sites_unicarbkb.csv",
          "filename": "sarscov1_proteoform_glycosylation_sites_unicarbkb.csv"
        }
      },
      {
        "mediatype": "log",
        "uri": {
          "access_time": "",
          "sha1_chksum": "",
          "uri": "https://data.glygen.org/ln2data/releases/data/v-1.12.3/reviewed/sarscov1_proteoform_glycosylation_sites_unicarbkb.1.log",
          "filename": "sarscov1_proteoform_glycosylation_sites_unicarbkb.1.log"
        }
      },
      {
        "mediatype": "log",
        "uri": {
          "access_time": "",
          "sha1_chksum": "",
          "uri": "https://data.glygen.org/ln2data/releases/data/v-1.12.3/reviewed/sarscov1_proteoform_glycosylation_sites_unicarbkb.2.log",
          "filename": "sarscov1_proteoform_glycosylation_sites_unicarbkb.2.log"
        }
      },
      {
        "mediatype": "csv",
        "uri": {
          "access_time": "",
          "sha1_chksum": "",
          "uri": "https://data.glygen.org/ln2data/releases/data/v-1.12.3/reviewed/sarscov1_proteoform_glycosylation_sites_unicarbkb.stat.csv",
          "filename": "sarscov1_proteoform_glycosylation_sites_unicarbkb.stat.csv"
        }
      }
    ]
  },
  "error_domain": {
    "empirical_error": {},
    "algorithmic_error": {}
  }
}