{
    "object_id": "https://biocomputeobject.org/GLY_001021/v-2.7.1",
    "spec_version": "1.3.0",
    "provenance_domain": {
        "name": "Mouse Glycosylation Citations (EMBL)",
        "version": "v-2.7.1",
        "license": "https://creativecommons.org/licenses/by/4.0/",
        "modified": "2024-05-17T11:24:22.227Z",
        "contributors": [
            {
                "name": "Jeet Vora",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "jeetvora@gwu.edu"
            },
            {
                "name": "Karina Martinez",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "karinamartinez@gwu.edu"
            },
            {
                "name": "Robel Kahsay",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "rykahsay@gwu.edu"
            },
            {
                "name": "Luke Johnson",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "luke.johnson@email.gwu.edu"
            },
            {
                "name": "Urnisha Bhuiyan",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "sbhuiyan28@email.gwu.edu"
            },
            {
                "name": "Cyrus Au-Yeung",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "cyruschauyeung@gwu.edu"
            },
            {
                "name": "Jingyue Wu",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "jingyue.wu@email.gwu.edu"
            },
            {
                "name": "Clement Potel",
                "affiliation": "EMBL",
                "email": "clement.potel@embl.de",
                "contribution": [
                    "contributedBy"
                ]
            },
            {
                "name": "Mira Burtscher",
                "affiliation": "EMBL",
                "email": "mira.burtscher@embl.de",
                "contribution": [
                    "contributedBy"
                ]
            },
            {
                "name": "Martin Garrido",
                "affiliation": "EMBL",
                "email": "martin.garrido@embl.de",
                "contribution": [
                    "contributedBy"
                ]
            },
            {
                "name": "Kate Warner",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "k.warner1@gwu.edu"
            }
        ]
    },
    "usability_domain": [
        "The Mouse Glycosylation Citations (EMBL) dataset contains publication information of the PMIDs present in the Mouse Glycosylation Sites (EMBL) dataset (https://data.glygen.org/GLY_000889)."
    ],
    "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": "Mus musculus",
                        "category_name": "species"
                    },
                    {
                        "category_value": "Proteoform",
                        "category_name": "molecule"
                    },
                    {
                        "category_value": "Glycosylation Sites",
                        "category_name": "tag"
                    },
                    {
                        "category_value": "csv",
                        "category_name": "file_type"
                    },
                    {
                        "category_value": "Reviewed",
                        "category_name": "status"
                    },
                    {
                        "category_value": "internal",
                        "category_name": "scope"
                    }
                ]
            }
        }
    ],
    "description_domain": {
        "keywords": [
            "Publications",
            "Citations",
            "PubMed",
            "glycosylation",
            "EMBL"
        ],
        "pipeline_steps": [
            {
                "name": "make-proteoform-dataset.py",
                "step_number": 1,
                "input_list": [],
                "description": "The script extracts the PMIDs from the dataset and uses the Entrez API to retrieve the publication details and extracts in a csv dataset",
                "prerequisite": [],
                "output_list": []
            }
        ]
    },
    "execution_domain": {
        "script": [
            {
                "uri": {
                    "uri": "https://github.com/glygener/glygen-backend-integration/blob/master/dataset-maker/make-proteoform-dataset.py",
                    "filename": "make-proteoform-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": {}
    },
    "parametric_domain": [
        {
            "value": ""
        }
    ],
    "io_domain": {
        "input_subdomain": [
            {
                "uri": {
                    "filename": "doi_citations.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.7.1/compiled/doi_citations.csv"
                }
            },
            {
                "uri": {
                    "filename": "mouse_proteoform_glycosylation_sites_embl.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.7.1/reviewed/mouse_proteoform_glycosylation_sites_embl.csv"
                }
            }
        ],
        "output_subdomain": [
            {
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.7.1/reviewed/mouse_proteoform_citations_glycosylation_sites_embl.csv",
                    "filename": "mouse_proteoform_citations_glycosylation_sites_embl.csv"
                },
                "mediatype": "csv"
            },
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.7.1/reviewed/mouse_proteoform_citations_glycosylation_sites_embl.stat.csv",
                    "filename": "mouse_proteoform_citations_glycosylation_sites_embl.stat.csv"
                }
            }
        ]
    },
    "etag": "005f19748af0a49408152b926476fa10"
}