{
    "object_id": "https://biocomputeobject.org/GLY_001191/v-2.12.1",
    "spec_version": "1.3.0",
    "provenance_domain": {
        "name": "Bovine Protein Citations (UniProtKB)",
        "version": "v-2.12.1",
        "license": "https://creativecommons.org/licenses/by/4.0/",
        "created": "2024-10-30T17:22:43",
        "modified": "2024-10-30T17:23:54.714Z",
        "contributors": [
            {
                "name": "Jeet Vora",
                "affiliation": "The George Washington University",
                "email": "jeetvora@gwu.edu",
                "contribution": [
                    "createdBy"
                ]
            },
            {
                "name": "Robel Kahsay",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "rykahsay@gwu.edu"
            },
            {
                "name": "Maria Martin",
                "affiliation": "EMBL-EBI-UniProt",
                "contribution": [
                    "contributedBy"
                ],
                "email": "martin@ebi.ac.uk"
            },
            {
                "name": "Preethi Vasudev",
                "affiliation": "EMBL-EBI-UniProt",
                "contribution": [
                    "contributedBy"
                ],
                "email": "pvasudev@ebi.ac.uk"
            },
            {
                "name": "Karina Martinez",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "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": "Jingyue Wu",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "jingyue.wu@email.gwu.edu"
            },
            {
                "name": "Kate Warner",
                "affiliation": "The George Washington University",
                "email": "k.warner1@gwu.edu",
                "contribution": [
                    "createdBy"
                ]
            }
        ]
    },
    "usability_domain": [
        "The Bovine Protein Citations dataset contains publication information for Bos taurus [taxid:9913] UniProtKB accessions that includes PMID, journal name, date and author and title. Publications of large scale studies, and high frequency publications pertaining to genome sequencing, cDNA, chromosomes have been filtered out. If you use this dataset please provide proper attribution to EMBL-EBI-UniProtKB and GlyGen."
    ],
    "description_domain": {
        "pipeline_steps": [
            {
                "name": "make-protein-dataset.py",
                "step_number": 1,
                "input_list": [],
                "description": "The script loads the UniProtKB publication data from the triplestore containing triples from nt file to retrieve the publication information associated with the UniProtKB accession and maps it to the UniProtKB canonical accessions. The script also filters out publications of large scale studies, and high frequency publications pertaining to DNA sequencing, cDNA, chromosomes using blacklisted PMIDs file.",
                "prerequisite": [],
                "output_list": []
            }
        ],
        "keywords": [
            "publications",
            "PMID",
            "journal",
            "citations",
            "authors",
            "UniProtKB",
            "PubMed"
        ]
    },
    "parametric_domain": [
        {
            "value": ""
        }
    ],
    "io_domain": {
        "output_subdomain": [
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.12.1/reviewed/bovine_protein_citations_uniprotkb.csv",
                    "filename": "bovine_protein_citations_uniprotkb.csv"
                }
            },
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.12.1/reviewed/bovine_protein_citations_uniprotkb.stat.csv",
                    "filename": "bovine_protein_citations_uniprotkb.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": "bovine_protein_function_uniprotkb.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.12.1/reviewed/bovine_protein_function_uniprotkb.csv"
                }
            },
            {
                "uri": {
                    "filename": "bovine_protein_ptm_annotation_uniprotkb.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.12.1/reviewed/bovine_protein_ptm_annotation_uniprotkb.csv"
                }
            },
            {
                "uri": {
                    "filename": "bovine_protein_site_annotation_uniprotkb.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.12.1/reviewed/bovine_protein_site_annotation_uniprotkb.csv"
                }
            },
            {
                "uri": {
                    "filename": "uniprot-proteome-bos-taurus.nt",
                    "uri": "https:/data.glygen.org/ln2downloads/ebi/2026_06_13/uniprot-proteome-bos-taurus.nt"
                }
            }
        ]
    },
    "execution_domain": {
        "script": [
            {
                "uri": {
                    "uri": "https://github.com/glygener/glygen-backend-integration/blob/master/dataset-maker/make-protein-dataset.py",
                    "filename": "make-protein-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": "Bos taurus",
                        "category_name": "species"
                    },
                    {
                        "category_value": "Protein",
                        "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": {},
    "etag": "bd280939653e9cb0d31bd85bf58a4b11"
}
