{
    "object_id": "https://biocomputeobject.org/GLY_000251/v-2.12.1",
    "spec_version": "1.3.0",
    "etag": "28e292adf98043cd7b5e562549fdf2ba",
    "provenance_domain": {
        "name": "Rat Proteoform Citations (UniCarbKB)",
        "version": "v-2.12.1",
        "license": "https://creativecommons.org/licenses/by/4.0/",
        "created": "2019-09-06T14:06:38.978102",
        "modified": "2024-12-10T18:47:54.650Z",
        "contributors": [
            {
                "name": "Jeet Vora",
                "affiliation": "The George Washington University",
                "email": "jeetvora@gwu.edu",
                "contribution": [
                    "createdBy"
                ]
            },
            {
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "name": "Robel Kahsay",
                "email": "rykahsay@gwu.edu"
            },
            {
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "name": "Karina Martinez",
                "email": "karinamartinez@gwu.edu"
            },
            {
                "contribution": [
                    "contributedBy"
                ],
                "name": "Not Applicable"
            },
            {
                "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 Rat Proteoform Citations (UniCarbKB) contains Rattus norvegicus [taxid:10116] publication information of glycoproteins from UniCarbKB. The script accesses the NCBI ENTREZ infrastructure to retrieve the publication information of the given PMID from UniCarbKB. The dataset is contributed by UniCarbKB and processed by GlyGen. If you use this dataset please provide proper attribution to UniCarbKB and GlyGen [UniCarbKB - https://academic.oup.com/nar/article/42/D1/D215/1052197 ,https://doi.org/10.1093/nar/gkt1128]. "
    ],
    "description_domain": {
        "pipeline_steps": [
            {
                "name": "make-proteoform-dataset.py",
                "step_number": 1,
                "input_list": [],
                "description": "The script parses the input file for PMIDs and accesses the NCBI Entrez infrastructure to download the publication information of the corresponding PMIDs. ",
                "prerequisite": [],
                "output_list": []
            }
        ],
        "keywords": [
            "PMID",
            "journal",
            "citations",
            "authors",
            "UniCarbKB",
            "PubMed"
        ]
    },
    "io_domain": {
        "output_subdomain": [
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.12.1/reviewed/rat_proteoform_citations_glycosylation_sites_unicarbkb.csv",
                    "filename": "rat_proteoform_citations_glycosylation_sites_unicarbkb.csv"
                }
            },
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.12.1/reviewed/rat_proteoform_citations_glycosylation_sites_unicarbkb.stat.csv",
                    "filename": "rat_proteoform_citations_glycosylation_sites_unicarbkb.stat.csv"
                }
            }
        ],
        "input_subdomain": [
            {
                "uri": {
                    "filename": "all_blacklisted_pmids.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.12.1/compiled/all_blacklisted_pmids.csv"
                }
            },
            {
                "uri": {
                    "filename": "doi_citations.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.12.1/compiled/doi_citations.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": "Rattus norvegicus",
                        "category_name": "species"
                    },
                    {
                        "category_value": "Proteoform",
                        "category_name": "molecule"
                    },
                    {
                        "category_value": "Protein Citations",
                        "category_name": "tag"
                    },
                    {
                        "category_value": "csv",
                        "category_name": "file_type"
                    },
                    {
                        "category_value": "Reviewed",
                        "category_name": "status"
                    },
                    {
                        "category_value": "internal",
                        "category_name": "scope"
                    },
                    {
                        "category_name": "file_category",
                        "category_value": "citations"
                    }
                ]
            }
        }
    ],
    "error_domain": {}
}