{
    "object_id": "https://biocomputeobject.org/GLY_000717/v-2.9.1",
    "spec_version": "1.3.0",
    "etag": "0aae9d7e2de9a10849a4175ada2cf08b",
    "provenance_domain": {
        "name": "Human O-Glucosylation Glycosylation Citations",
        "version": "v-2.9.1",
        "license": "https://creativecommons.org/licenses/by/4.0/",
        "created": "2023-02-28T15:39:52.735Z",
        "modified": "2025-03-19T17:33:33.763Z",
        "contributors": [
            {
                "name": "Jeet Vora",
                "affiliation": "The George Washington University",
                "email": "jeetvora@gwu.edu",
                "contribution": [
                    "createdBy",
                    "curatedBy"
                ]
            },
            {
                "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": "Daniel Williamson",
                "contribution": [
                    "authoredBy",
                    "contributedBy"
                ],
                "affiliation": "CCRC, UGA",
                "email": "daniel.williamson25@uga.edu"
            },
            {
                "name": "Robert S Haltiwanger",
                "contribution": [
                    "createdAt",
                    "contributedBy"
                ],
                "affiliation": "CCRC, UGA",
                "email": "rhalti@uga.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": "Kate Warner",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "k.warner1@gwu.edu"
            }
        ]
    },
    "usability_domain": [
        "The O-Glucosylation Glycosylation Citations datasets contain details of the publications present in the Human O-Glucosylation Glycosylation Sites  experimental and predicted datasets https://data.glygen.org/GLY_000716, contributed by Daniel Williamson from Bob Haltiwanger's group. The data from this paper has been split into two datasets, one which contains details of the publications present in the experimental dataset (human_proteoform_citations_glycosylation_sites_o_gluc.csv) and the other contains publications present in the predicted dataset (human_proteoform_citations_glycosylation_sites_o_gluc_predicted.csv)"
    ],
    "description_domain": {
        "pipeline_steps": [
            {
                "name": "make-proteoform-dataset.py",
                "step_number": 1,
                "input_list": [],
                "description": "The script extracts the PMID from the dataset and uses the Entrez API to retrieve the publication details and extracts in a csv dataset. ",
                "prerequisite": [],
                "output_list": []
            }
        ],
        "keywords": [
            "O-Gluc",
            "O-Glucosylation",
            "Publications",
            "Citations",
            "PubMed"
        ]
    },
    "parametric_domain": [
        {
            "value": ""
        }
    ],
    "io_domain": {
        "output_subdomain": [
            {
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.9.1/reviewed/human_proteoform_citations_glycosylation_sites_o_gluc.csv",
                    "filename": "human_proteoform_citations_glycosylation_sites_o_gluc.csv"
                },
                "mediatype": "csv"
            },
            {
                "mediatype": "csv",
                "uri": {
                    "filename": "human_proteoform_citations_glycosylation_sites_o_gluc_predicted.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.9.1/reviewed/human_proteoform_citations_glycosylation_sites_o_gluc_predicted.csv"
                }
            },
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.9.1/reviewed/human_proteoform_citations_glycosylation_sites_o_gluc.stat.csv",
                    "filename": "human_proteoform_citations_glycosylation_sites_o_gluc.stat.csv"
                }
            },
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.9.1/reviewed/human_proteoform_citations_glycosylation_sites_o_gluc_predicted.stat.csv",
                    "filename": "human_proteoform_citations_glycosylation_sites_o_gluc_predicted.stat.csv"
                }
            }
        ],
        "input_subdomain": [
            {
                "uri": {
                    "filename": "all_blacklisted_pmids.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.9.1/compiled/all_blacklisted_pmids.csv"
                }
            },
            {
                "uri": {
                    "filename": "human_proteoform_glycosylation_sites_o_gluc.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.9.1/reviewed/human_proteoform_glycosylation_sites_o_gluc.csv"
                }
            },
            {
                "uri": {
                    "filename": "all_blacklisted_pmids.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.9.1/compiled/all_blacklisted_pmids.csv"
                }
            },
            {
                "uri": {
                    "filename": "human_proteoform_glycosylation_sites_o_gluc_predicted.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.9.1/reviewed/human_proteoform_glycosylation_sites_o_gluc_predicted.csv"
                }
            }
        ]
    },
    "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": {}
    },
    "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": "Proteoform",
                        "category_name": "molecule"
                    },
                    {
                        "category_value": "csv",
                        "category_name": "file_type"
                    },
                    {
                        "category_value": "Reviewed",
                        "category_name": "status"
                    },
                    {
                        "category_value": "internal",
                        "category_name": "scope"
                    }
                ]
            }
        }
    ],
    "error_domain": {}
}