{
    "object_id": "https://biocomputeobject.org/GLY_000081/v-1.0.13",
    "license": "https://creativecommons.org/licenses/by/4.0/",
    "contributors": [
        {
            "orcid": "",
            "affiliation": "The George Washington University",
            "contribution": [
                "createdBy"
            ],
            "name": "Jeet Vora",
            "email": "jeetvora@gwu.edu"
        },
        {
            "orcid": "",
            "affiliation": "The George Washington University",
            "contribution": [
                "createdBy"
            ],
            "name": "Rahi Navelkar",
            "email": "rsn13@gwu.edu"
        },
        {
            "orcid": "",
            "affiliation": "The George Washington University",
            "contribution": [],
            "name": "Robel Kahsay",
            "email": "rykahsay@gwu.edu"
        },
        {
            "orcid": "",
            "affiliation": "EMBL-EBI-UniProt",
            "contribution": [
                "contributedBy"
            ],
            "name": "Maria Martin",
            "email": "martin@ebi.ac.uk"
        },
        {
            "orcid": "",
            "affiliation": "EMBL-EBI-UniProt",
            "contribution": [
                "contributedBy"
            ],
            "name": "Leyla Garcia",
            "email": "ljgarcia@ebi.ac.uk"
        },
        {
            "orcid": "",
            "affiliation": "EMBL-EBI-UniProt",
            "contribution": [
                "contributedBy"
            ],
            "name": "Preethi Vasudev",
            "email": "pvasudev@ebi.ac.uk"
        }
    ],
    "created": "2019-01-23T17:21:05.950210",
    "spec_version": "1.3.0",
    "modified": "2019-02-21T12:10:34.108618",
    "extension_domain": [
        {
            "extension_schema": "http://www.w3id.org/biocompute/extension_domain/1.1.0/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": "Reviewed",
                        "category_name": "status"
                    }
                ]
            }
        }
    ],
    "io_domain": {
        "input_subdomain": [
            {
                "uri": {
                    "uri": "http://data.glygen.org/datasets/source/uniprot-proteome-9606-homo-sapiens-2018-11.nt",
                    "filename": "uniprot-proteome-9606-homo-sapiens-2018-11.nt"
                }
            }
        ],
        "output_subdomain": [
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "http://data.glygen.org/datasets/reviewed/human_protein_genelocus.csv",
                    "filename": "human_protein_genelocus.csv"
                }
            },
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "http://data.glygen.org/datasets/reviewed/human_protein_genelocus.stat.csv",
                    "filename": "human_protein_genelocus.stat.csv"
                }
            }
        ]
    },
    "provenance_domain": {
        "license": "https://creativecommons.org/licenses/by/4.0/",
        "name": "Human Gene Locus",
        "contributors": [
            {
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "name": "Jeet Vora",
                "email": "jeetvora@gwu.edu"
            },
            {
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "name": "Rahi Navelkar",
                "email": "rsn13@gwu.edu"
            },
            {
                "affiliation": "The George Washington University",
                "contribution": [],
                "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": "Leyla Garcia",
                "email": "ljgarcia@ebi.ac.uk"
            },
            {
                "affiliation": "EMBL-EBI-UniProt",
                "contribution": [
                    "contributedBy"
                ],
                "name": "Preethi Vasudev",
                "email": "pvasudev@ebi.ac.uk"
            }
        ],
        "created": "2019-01-23T17:21:05.950210",
        "modified": "2019-03-16T17:06:24.097514",
        "version": "v-1.0.13"
    },
    "description_domain": {
        "keywords": [
            "species: homo sapiens",
            "dataset: genelocus",
            "category: protein",
            "gene locus"
        ],
        "pipeline_steps": [
            {
                "name": "extract-from-endpoint.py",
                "step_number": 1,
                "version": "1.0",
                "input_list": [
                    {
                        "uri": "http://data.glygen.org/datasets/source/uniprot-proteome-9606-homo-sapiens-2018-11.nt",
                        "filename": "uniprot-proteome-9606-homo-sapiens-2018-11.nt"
                    }
                ],
                "output_list": [
                    {
                        "uri": "http://data.glygen.org/GLYDS000081",
                        "filename": "human_protein_genelocus.csv"
                    }
                ],
                "prerequisite": [],
                "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"
            }
        ]
    },
    "execution_domain": {
        "external_data_endpoints": [],
        "environment_variables": {},
        "script_driver": "manual",
        "software_prerequisites": [
            {
                "version": "2.7.5",
                "name": "Python",
                "uri": {
                    "uri": "https://www.python.org/download/releases/2.7.5/"
                }
            }
        ],
        "script": [
            {
                "uri": {
                    "uri": "XXXXXXXXXXXXXXXXXXX",
                    "filename": "extract-from-endpoint.py"
                }
            }
        ]
    },
    "version": "1.0.5",
    "error_domain": {
        "algorithmic_error": {},
        "emperical_error": {}
    },
    "parametric_domain": [
        {
            "step": "",
            "value": "",
            "param": ""
        }
    ],
    "usability_domain": [
        "Human (taxid:9606) Gene Locus contains UniProtKB canonical accessions mapped to the corresponding ENSEMBL Gene ID, chromosome ID along with the gene's start and end position on the chromosome",
        "The dataset is derived from 2018-11 UniProt release. The dataset is contributed by EMBL-EB-UniProt and processed by GlyGen. If you use this dataset please provide proper attribution to EMBL-EB-UniProt and GlyGen"
    ],
    "etag": "9f32cb1a4aa5adfbf73ecfe5083455b6",
    "_id": {
        "$oid": "5c48e8d1f5c47b65ae4b5d3c"
    },
    "name": "Human Gene Locus"
}