{
    "object_id": "https://biocomputeobject.org/GLY_000751/v-2.5.1",
    "etag": "f21bca192305ceb488c866d9aac37a31",
    "spec_version": "1.3.0",
    "provenance_domain": {
        "name": "Yeast Gene Locus (Ensembl)",
        "version": "v-2.5.1",
        "created": "2020-04-10T11:29:22.824792",
        "modified": "2024-05-17T11:24:11.820Z",
        "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": "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": "https://creativecommons.org/licenses/by/4.0/"
    },
    "usability_domain": [
        "The Yeast Gene Locus (Ensembl) contains Saccharomyces cerevisiae S288C [taxid:559292] UniProtKB canonical accessions mapped to the corresponding ensembl Gene ID, chromosome ID along with the gene's start and end position on the chromosome. If you use this dataset please provide proper attribution to EMBL-EBI-UniProtKB and GlyGen."
    ],
    "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": "Gene Coordinates",
                        "category_name": "tag"
                    },
                    {
                        "category_value": "csv",
                        "category_name": "file_type"
                    },
                    {
                        "category_value": "Reviewed",
                        "category_name": "status"
                    },
                    {
                        "category_value": "internal",
                        "category_name": "scope"
                    }
                ]
            }
        }
    ],
    "description_domain": {
        "keywords": [
            "gene coordinates",
            "chromosome",
            "start position",
            "end position",
            "Ensembl",
            "strand"
        ],
        "pipeline_steps": [
            {
                "step_number": 1,
                "name": "make-protein-dataset.py",
                "description": "The script loads the UniProtKB canonical accessions that are mapped to ENSEMBL Gene identifiers from the triplestore containing triples from nt file and extracts the data as a csv dataset",
                "version": "",
                "prerequisite": [],
                "input_list": [],
                "output_list": []
            }
        ]
    },
    "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": {}
    },
    "parametric_domain": [
        {
            "param": "",
            "value": "",
            "step": ""
        }
    ],
    "io_domain": {
        "input_subdomain": [
            {
                "uri": {
                    "filename": "yeast_protein_masterlist.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.5.1/reviewed/yeast_protein_masterlist.csv"
                }
            },
            {
                "uri": {
                    "filename": "uniprot-proteome-saccharomyces-cerevisiae.nt",
                    "uri": "https:/data.glygen.org/ln2downloads/ebi/2024_04_08/uniprot-proteome-saccharomyces-cerevisiae.nt"
                }
            }
        ],
        "output_subdomain": [
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.5.1/reviewed/yeast_protein_genelocus.csv",
                    "filename": "yeast_protein_genelocus.csv"
                }
            },
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.5.1/reviewed/yeast_protein_genelocus.stat.csv",
                    "filename": "yeast_protein_genelocus.stat.csv"
                }
            }
        ]
    },
    "error_domain": {
        "empirical_error": {},
        "algorithmic_error": {}
    }
}