{
    "object_id": "https://biocomputeobject.org/GLY_000804/v-2.9.1",
    "spec_version": "1.3.0",
    "etag": "342f8c3e053afc9f28904dfbf8b7ea90",
    "provenance_domain": {
        "name": "Yeast Proteoform Phosphorylation Citations (UniProtKB)",
        "version": "v-2.9.1",
        "license": "https://creativecommons.org/licenses/by/4.0/",
        "created": "2022-09-02T13:32:35.505543",
        "modified": "2024-12-10T21:35:22.768Z",
        "contributors": [
            {
                "email": "jeetvora@gwu.edu",
                "contribution": [
                    "createdBy"
                ]
            },
            {
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "name": "Karina Martinez",
                "email": "karinamartinez@gwu.edu"
            },
            {
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "name": "Robel Kahsay",
                "email": "rykahsay@gwu.edu"
            },
            {
                "name": "Jingyue Wu",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "jingyue.wu@email.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"
            }
        ]
    },
    "usability_domain": [
        "The Yeast Proteoform Phosphorylation Citations (UniProtKB) dataset contains Saccharomyces cerevisiae S288C [taxid:559292] publication information of phosphorylated glycoproteins from UniProtKB. The script accesses the NCBI ENTREZ infrastructure to retrieve the publication information of the given PMID from UniProtKB."
    ],
    "description_domain": {
        "pipeline_steps": [
            {
                "step_number": 1,
                "name": "make-proteoform-dataset.py",
                "description": "The script parses the input file for PMIDs and accesses the NCBI Entrez infrastructure to download the publication information of the corresponding PMIDs",
                "input_list": [],
                "prerequisite": [],
                "output_list": []
            }
        ],
        "keywords": [
            "publications",
            "PMID",
            "journal",
            "citations",
            "authors",
            "UniProtKB",
            "PubMed",
            "Phosphorylation"
        ]
    },
    "parametric_domain": [
        {
            "value": ""
        }
    ],
    "io_domain": {
        "output_subdomain": [
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.9.1/reviewed/yeast_proteoform_citations_phosphorylation_sites_uniprotkb.csv",
                    "filename": "yeast_proteoform_citations_phosphorylation_sites_uniprotkb.csv"
                }
            },
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.9.1/reviewed/yeast_proteoform_citations_phosphorylation_sites_uniprotkb.stat.csv",
                    "filename": "yeast_proteoform_citations_phosphorylation_sites_uniprotkb.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": "yeast_proteoform_phosphorylation_sites_uniprotkb.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.9.1/reviewed/yeast_proteoform_phosphorylation_sites_uniprotkb.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": "Saccharomyces cerevisiae S288C",
                        "category_name": "species"
                    },
                    {
                        "category_value": "Protein",
                        "category_name": "molecule"
                    },
                    {
                        "category_value": "Phosphorylation Citations UniProtKB",
                        "category_name": "tag"
                    },
                    {
                        "category_value": "csv",
                        "category_name": "file_type"
                    },
                    {
                        "category_value": "Reviewed",
                        "category_name": "status"
                    },
                    {
                        "category_value": "internal",
                        "category_name": "scope"
                    }
                ]
            }
        }
    ],
    "error_domain": {}
}