{
    "object_id": "https://biocomputeobject.org/GLY_000245/v-1.12.1",
    "spec_version": "1.3.0",
    "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": "Rattus norvegicus",
                        "category_name": "species"
                    },
                    {
                        "category_value": "Protein",
                        "category_name": "molecule"
                    },
                    {
                        "category_value": "Pdb cross-references",
                        "category_name": "tags"
                    },
                    {
                        "category_value": "csv",
                        "category_name": "file_type"
                    },
                    {
                        "category_value": "Reviewed",
                        "category_name": "status"
                    },
                    {
                        "category_value": "internal",
                        "category_name": "scope"
                    }
                ]
            }
        }
    ],
    "provenance_domain": {
        "license": "https://creativecommons.org/licenses/by/4.0/",
        "name": "Rat UniProtKB Xref PDB",
        "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": [
                    "createdBy"
                ],
                "name": "Robel Kahsay",
                "email": "rykahsay@gwu.edu"
            },
            {
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "name": "Xavier Holmes",
                "email": "xavierh@email.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"
            },
            {
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "name": "Amanda Bell",
                "email": "amandab2140@gwu.edu"
            },
            {
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "name": "Jonathon Keeney",
                "email": "keeneyjg@gwu.edu"
            },
            {
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "name": "Millicent Quartey",
                "email": "mquartey@email.gwu.edu"
            }
        ],
        "created": "2019-09-06T14:06:38.971888",
        "modified": "2022-01-19T13:01:25.317411",
        "version": "v-1.12.1"
    },
    "description_domain": {
        "keywords": [
            "protein 3D structure",
            "RCSB PDB",
            "cross-reference"
        ],
        "pipeline_steps": [
            {
                "name": "make-protein-dataset.py",
                "step_number": 1,
                "version": "",
                "input_list": [
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-1.12.1/reviewed/rat_protein_masterlist.csv",
                        "filename": "rat_protein_masterlist.csv"
                    },
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-1.12.1/source/uniprot-proteome-rattus-norvegicus.nt",
                        "filename": "uniprot-proteome-rattus-norvegicus.nt"
                    }
                ],
                "output_list": [
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-1.12.1/reviewed/rat_protein_xref_pdb.csv",
                        "filename": "rat_protein_xref_pdb.csv"
                    }
                ],
                "prerequisite": [],
                "description": "The script loads the UniProtKB canonical accessions that are cross-referenced to RCSB PDB database accessions/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": "https://github.com/GW-HIVE/glygen-backend-integration/blob/master/pipeline/integrator/make-protein-dataset.py",
                    "filename": "make-protein-dataset.py"
                }
            }
        ]
    },
    "usability_domain": [
        "The Rat UniProtKB Xref PDB contains rat [taxid:10116] UniProtKB canonical accessions cross-referenced to PDB database accessions/identifiers.  The dataset is derived from 2019-09 UniProtKB release. If you use this dataset please provide proper attribution to EMBL-EBI-UniProtKB and GlyGen. ",
        "The Protein Data Bank is a database for the three-dimensional structural data of large biological molecules, such as proteins and nucleic acids. "
    ],
    "etag": "ce5b3168295dff0b17f8faff8c8a26b3",
    "io_domain": {
        "input_subdomain": [
            {
                "uri": {
                    "uri": "https:/data.glygen.org/ln2data/releases/data/v-x.x.x/reviewed/rat_protein_ntdata.nt",
                    "filename": "rat_protein_uniprot-proteome-rattus-norvegicus.ntcsv"
                }
            },
            {
                "uri": {
                    "uri": "https:/data.glygen.org/ln2downloads/ebi/v-x.x.x/uniprot-proteome-rattus-norvegicus.nt",
                    "filename": "uniprot-proteome-rattus-norvegicus.nt"
                }
            }
        ],
        "output_subdomain": [
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2data/releases/data/v-x.x.x/reviewed/rat_protein_xref_pdb.csv",
                    "filename": "rat_protein_xref_pdb.csv"
                }
            }
        ]
    }
}
