{
    "object_id": "https://biocomputeobject.org/GLY_001059/v-2.8.1",
    "spec_version": "1.3.0",
    "etag": "2c862d4c10ed4f28d4fa2374ee34267a",
    "provenance_domain": {
        "name": "Arabidopsis Gene Locus (Ensembl)",
        "version": "v-2.8.1",
        "license": "https://creativecommons.org/licenses/by/4.0/",
        "modified": "2024-07-08T15:17:44.182Z",
        "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",
                "contribution": [
                    "createdBy"
                ],
                "email": "k.warner1@gwu.edu"
            }
        ]
    },
    "usability_domain": [
        "The Arabidopsis Gene Locus (Ensembl) contains Arabidopsis thaliana [taxid:3702] 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."
    ],
    "description_domain": {
        "pipeline_steps": [
            {
                "name": "make-protein-dataset.py",
                "step_number": 1,
                "input_list": [],
                "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",
                "prerequisite": [],
                "output_list": []
            }
        ],
        "keywords": [
            "gene coordinates",
            "chromosome",
            "start position",
            "end position",
            "Ensembl",
            "strand"
        ]
    },
    "parametric_domain": [
        {
            "value": ""
        }
    ],
    "io_domain": {
        "output_subdomain": [
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.8.1/reviewed/arabidopsis_protein_genelocus.csv",
                    "filename": "arabidopsis_protein_genelocus.csv"
                }
            },
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.8.1/reviewed/arabidopsis_protein_genelocus.stat.csv",
                    "filename": "arabidopsis_protein_genelocus.stat.csv"
                }
            }
        ],
        "input_subdomain": [
            {
                "uri": {
                    "filename": "arabidopsis_protein_masterlist.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.8.1/reviewed/arabidopsis_protein_masterlist.csv"
                }
            },
            {
                "uri": {
                    "filename": "uniprot-proteome-arabidopsis-thaliana.nt",
                    "uri": "https:/data.glygen.org/ln2downloads/ebi/2025_01_17/uniprot-proteome-arabidopsis-thaliana.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": "Arabidopsis thaliana",
                        "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"
                    }
                ]
            }
        }
    ],
    "error_domain": {}
}