{
    "object_id": "https://biocomputeobject.org/GLY_000799/v-2.9.1",
    "etag": "9aae0abfe43d3933d7399f272531971b",
    "spec_version": "1.3.0",
    "provenance_domain": {
        "name": "Yeast Proteoform Citations (GlyConnect)",
        "version": "v-2.9.1",
        "created": "2022-09-02T10:28:05.203794",
        "modified": "2024-05-17T11:24:13.717Z",
        "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": "GlyConnect",
                "contribution": [
                    "contributedBy"
                ],
                "name": "Frederique Lisacek",
                "email": "frederique.lisacek@sib.swiss"
            },
            {
                "affiliation": "GlyConnect",
                "contribution": [
                    "contributedBy"
                ],
                "name": "Julien Mariethoz",
                "email": "julien.mariethoz@sib.swiss"
            },
            {
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "name": "Karina Martinez",
                "email": "karinamartinez@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"
            }
        ],
        "license": "httpss://creativecommons.org/licenses/by/4.0/"
    },
    "usability_domain": [
        "The Yeast Proteoform Citations (GlyConnect) dataset provides citation details corresponding to the PubMed ID (PMID) from file httpss://data.glygen.org/GLYDS000329. The script accesses the NCBI ENTREZ infrastructure to retrieve the publication information of the given PMID. The dataset is contributed by GlyGen and processed by GlyGen. If you use this dataset please provide proper attribution to GlyConnect (httpss://glyconnect.expasy.org/)."
    ],
    "description_domain": {
        "keywords": [
            "PMID",
            "journal",
            "citations",
            "authors",
            "GlyConnect",
            "PubMed"
        ],
        "platform": [],
        "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",
                "version": "",
                "prerequisite": [],
                "input_list": [],
                "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": {}
    },
    "io_domain": {
        "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_glycosylation_sites_glyconnect.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.9.1/reviewed/yeast_proteoform_glycosylation_sites_glyconnect.csv"
                }
            }
        ],
        "output_subdomain": [
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.9.1/reviewed/yeast_proteoform_citations_glycosylation_sites_glyconnect.csv",
                    "filename": "yeast_proteoform_citations_glycosylation_sites_glyconnect.csv"
                }
            },
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.9.1/reviewed/yeast_proteoform_citations_glycosylation_sites_glyconnect.stat.csv",
                    "filename": "yeast_proteoform_citations_glycosylation_sites_glyconnect.stat.csv"
                }
            }
        ]
    },
    "parametric_domain": [
        {
            "param": "",
            "value": "",
            "step": ""
        }
    ],
    "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": "Saccharomyces cerevisiae S288C",
                        "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"
                    }
                ]
            }
        }
    ]
}