{
    "object_id": "https://biocomputeobject.org/GLY_000138/v-2.5.1",
    "spec_version": "1.3.0",
    "etag": "db59554ba1b04b3fd87b5cc44b499a62",
    "provenance_domain": {
        "name": "Human Proteoform Citations (UniCarbKB)",
        "version": "v-2.5.1",
        "created": "2019-04-16T16:31:37.511217",
        "modified": "2024-05-17T11:23:52.401Z",
        "embargo": {
            "start_time": "",
            "end_time": ""
        },
        "contributors": [
            {
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "name": "Jeet Vora",
                "email": "jeetvora@gwu.edu"
            },
            {
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "name": "Robel Kahsay",
                "email": "rykahsay@gwu.edu"
            },
            {
                "affiliation": "EMBL-EBI-UniProt",
                "contribution": [
                    "contributedBy"
                ],
                "name": "Maria Martin",
                "email": "martin@ebi.ac.uk"
            },
            {
                "affiliation": "EMBL-EBI-UniProt",
                "contribution": [
                    "contributedBy"
                ],
                "name": "Preethi Vasudev",
                "email": "pvasudev@ebi.ac.uk"
            },
            {
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "name": "Karina Martinez",
                "email": "karinamartinez@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": "Kate Warner",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "k.warner1@gwu.edu"
            }
        ],
        "license": "https://creativecommons.org/licenses/by/4.0/"
    },
    "usability_domain": [
        "The Human Proteoform Citations (UniCarbKB) contains human [taxid:9606] 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": {
        "keywords": [
            "PMID",
            "journal",
            "citations",
            "authors",
            "UniCarbKB",
            "PubMed"
        ],
        "platform": [],
        "pipeline_steps": [
            {
                "name": "make-proteoform-dataset.py",
                "step_number": 1,
                "version": "",
                "input_list": [],
                "output_list": [],
                "prerequisite": [],
                "description": "The script parses the input file for PMIDs and accesses the NCBI Entrez infrastructure to download the publication information of the corresponding PMIDs. "
            }
        ]
    },
    "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": {}
    },
    "io_domain": {
        "input_subdomain": [
            {
                "uri": {
                    "filename": "doi_citations.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.5.1/compiled/doi_citations.csv"
                }
            },
            {
                "uri": {
                    "filename": "human_proteoform_glycosylation_sites_unicarbkb.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.5.1/reviewed/human_proteoform_glycosylation_sites_unicarbkb.csv"
                }
            }
        ],
        "output_subdomain": [
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.5.1/reviewed/human_proteoform_citations_glycosylation_sites_unicarbkb.csv",
                    "filename": "human_proteoform_citations_glycosylation_sites_unicarbkb.csv"
                }
            },
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.5.1/reviewed/human_proteoform_citations_glycosylation_sites_unicarbkb.stat.csv",
                    "filename": "human_proteoform_citations_glycosylation_sites_unicarbkb.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": "Homo sapiens",
                        "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"
                    }
                ]
            }
        }
    ]
}