{
    "object_id": "https://biocomputeobject.org/GLY_000389/v-2.0.3",
    "spec_version": "1.3.0",
    "extension_domain": [
        {
            "extension_schema": "http://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": "Rattus norvegicus",
                        "category_name": "species"
                    },
                    {
                        "category_value": "Protein",
                        "category_name": "molecule"
                    },
                    {
                        "category_value": "Protein Information",
                        "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 Gene Symbols (NCBI RefSeq)",
        "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": "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"
            },
            {
                "affiliation": "",
                "contribution": [
                    "contributedBy"
                ],
                "name": "Not Applicable",
                "email": ""
            }
        ],
        "created": "2020-03-15T12:52:59.660127",
        "modified": "2022-09-26T09:05:27.653481",
        "version": "v-2.0.3"
    },
    "description_domain": {
        "keywords": [
            "protein information",
            "protein length",
            "NCBI RefSeq",
            "protein summary"
        ],
        "pipeline_steps": [
            {
                "name": "make-protein-dataset.py",
                "step_number": 1,
                "version": "",
                "input_list": [
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-2.0.3/reviewed/rat_protein_xref_refseq.csv",
                        "filename": "rat_protein_xref_refseq.csv"
                    },
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-2.0.3/source/refseq_protein_all_10116.gpff",
                        "filename": "refseq_protein_all_10116.gpff"
                    }
                ],
                "output_list": [
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-2.0.3/reviewed/rat_protein_genenames_refseq.csv",
                        "filename": "rat_protein_genenames_refseq.csv"
                    }
                ],
                "prerequisite": [],
                "description": "The script parses the NCBI RefSeq gpff files to extract the gene symbols/names and then maps the NCBI RefSeq accesion to the UniProtKB Canonical accessions"
            }
        ]
    },
    "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/glygener/glygen-backend-integration/blob/master/dataset-maker/make-protein-dataset.py",
                    "filename": "make-protein-dataset.py"
                }
            }
        ]
    },
    "error_domain": {
        "empirical_error": {},
        "algorithmic_error": {}
    },
    "parametric_domain": [
        {
            "step": "",
            "param": "",
            "value": ""
        }
    ],
    "usability_domain": [
        "The Rat Gene Symbols (NCBI RefSeq) dataset contains rat [taxid:10116] gene symbols/names (primary gene name and gene synonyms)  from NCBI RefSeq database.  The dataset is derived from NCBI RefSeq Release 96, September 9, 2019. If you use this dataset please provide proper attribution to NCBI RefSeq and GlyGen"
    ],
    "etag": "39ae55b954ded0a3261eaf0cfa421e35",
    "io_domain": {
        "input_subdomain": [
            {
                "uri": {
                    "filename": "rat_protein_xref_refseq.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.0.3/reviewed/rat_protein_xref_refseq.csv"
                }
            }
        ],
        "output_subdomain": [
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.0.3/reviewed/rat_protein_genenames_refseq.csv",
                    "filename": "rat_protein_genenames_refseq.csv"
                }
            },
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.0.3/reviewed/rat_protein_genenames_refseq.stat.csv",
                    "filename": "rat_protein_genenames_refseq.stat.csv"
                }
            }
        ]
    }
}