{
    "object_id": "https://biocomputeobject.org/GLY_000615/v-2.2.1",
    "spec_version": "1.3.0",
    "etag": "4dde22880492a71bc922de20ea64f43f",
    "provenance_domain": {
        "name": "SARS-CoV2 NCBI Protein Linkouts",
        "version": "v-2.2.1",
        "created": "2021-10-15T15:18:43.701633",
        "modified": "2023-03-31T18:52:18.869Z",
        "embargo": {
            "start_time": "",
            "end_time": ""
        },
        "contributors": [
            {
                "affiliation": "",
                "contribution": [
                    "createdBy"
                ],
                "name": "",
                "email": "jeetvora@gwu.edu"
            },
            {
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "name": "Karina Martinez",
                "email": "karinamartinez@gwu.edu"
            },
            {
                "name": "Jingyue Wu",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "jingyue.wu@email.gwu.edu"
            }
        ],
        "license": "https://creativecommons.org/licenses/by/4.0/"
    },
    "usability_domain": [
        "The The SARS coronavirus (SARS-CoV-2 or 2019-nCoV) NCBI Protein Linkouts contains sarscov2 [taxid:2697049] NCBI RefSeq accessions with their corresponding GlyGen protein detail page URLs. The dataset is formatted as per NCBI non-bib. linkouts guidelines where 10277 is the GlyGen's provider ID.",
        "The RefSeq accessions are The dataset is derived from NCBI RefSeq Release 96, September 9, 2019"
    ],
    "description_domain": {
        "keywords": [
            "NCBI Protein Linkouts",
            "RefSeq"
        ],
        "platform": [],
        "pipeline_steps": [
            {
                "name": "make-species-agnostic-dataset.py",
                "step_number": 1,
                "version": "",
                "input_list": [],
                "output_list": [],
                "prerequisite": [],
                "description": "The script loads the RefSeq accessions, RefSeq protein name and based on the RefSeq-UniprotKB accession mapping generates the GlyGen protein detail page URLs. The script also enters the other fields as defined by the NCBI Protein Linkouts to generate a csv dataset"
            }
        ]
    },
    "execution_domain": {
        "script": [
            {
                "uri": {
                    "uri": "https://github.com/glygener/glygen-backend-integration/blob/master/dataset-maker/make-protein-dataset.py",
                    "filename": "make-protein-dataset.py"
                }
            }
        ],
        "script_driver": "manual",
        "software_prerequisites": [
            {
                "version": "2.7.5",
                "name": "Python",
                "uri": {
                    "uri": "https://www.python.org/download/releases/2.7.5/"
                }
            }
        ],
        "external_data_endpoints": [],
        "environment_variables": {}
    },
    "io_domain": {
        "input_subdomain": [
            {
                "uri": {
                    "filename": "sarscov2_protein_proteinnames_refseq.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.2.1/reviewed/sarscov2_protein_proteinnames_refseq.csv"
                }
            },
            {
                "uri": {
                    "filename": "sarscov2_protein_xref_refseq.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.2.1/reviewed/sarscov2_protein_xref_refseq.csv"
                }
            }
        ],
        "output_subdomain": [
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.2.1/reviewed/sarscov2_protein_ncbi_linkouts.csv",
                    "filename": "sarscov2_protein_ncbi_linkouts.csv"
                }
            },
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.2.1/reviewed/sarscov2_protein_ncbi_linkouts.stat.csv",
                    "filename": "sarscov2_protein_ncbi_linkouts.stat.csv"
                }
            }
        ]
    },
    "parametric_domain": [
        {
            "step": "",
            "param": "",
            "value": ""
        }
    ],
    "error_domain": {
        "empirical_error": {},
        "algorithmic_error": {}
    },
    "extension_domain": [
        {
            "extension_schema": "https://www.w3id.org/biocompute/extension_domain/1.2.0/dataset/dataset_extension.json",
            "dataset_extension": {
                "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": "Severe acute respiratory syndrome coronavirus 2",
                        "category_name": "species"
                    },
                    {
                        "category_value": "Protein",
                        "category_name": "molecule"
                    },
                    {
                        "category_value": "NCBI Gene Linkouts",
                        "category_name": "tag"
                    },
                    {
                        "category_value": "csv",
                        "category_name": "file_type"
                    },
                    {
                        "category_value": "Reviewed",
                        "category_name": "status"
                    },
                    {
                        "category_value": "internal",
                        "category_name": "scope"
                    }
                ]
            }
        }
    ]
}