{
    "object_id": "https://biocomputeobject.org/GLY_000022/v-2.3.1",
    "spec_version": "1.3.0",
    "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": "Mus musculus",
                        "category_name": "species"
                    },
                    {
                        "category_value": "Protein",
                        "category_name": "molecule"
                    },
                    {
                        "category_value": "Protein Information",
                        "category_name": "tag"
                    },
                    {
                        "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": "Mouse Protein Information (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": "Robel Kahsay",
                "email": "rykahsay@gwu.edu"
            },
            {
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "name": "Karina Martinez",
                "email": "karinamartinez@gwu.edu"
            },
            {
                "affiliation": "",
                "contribution": [
                    "contributedBy"
                ],
                "name": "Not Applicable",
                "email": ""
            },
            {
                "name": "Jingyue Wu",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "jingyue.wu@email.gwu.edu"
            }
        ],
        "created": "2019-01-23T17:21:05.888176",
        "modified": "2022-09-26T08:48:18.844146",
        "version": "v-2.3.1"
    },
    "description_domain": {
        "keywords": [
            "protein information",
            "protein length",
            "protein summary",
            "NCBI RefSeq"
        ],
        "pipeline_steps": [
            {
                "name": "make-protein-dataset.py",
                "step_number": 1,
                "version": "",
                "input_list": [],
                "output_list": [],
                "prerequisite": [],
                "description": "The script parses the NCBI RefSeq gpff files to extract the accession, length, protein summary 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 Mouse Protein Information (NCBI RefSeq) dataset contains mouse [taxid:10090] protein information such as protein accession, name, length and protein function summary 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": "859e036e865edd625cf0814e205ee396",
    "io_domain": {
        "input_subdomain": [
            {
                "uri": {
                    "filename": "mouse_protein_xref_refseq.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.3.1/reviewed/mouse_protein_xref_refseq.csv"
                }
            },
            {
                "uri": {
                    "filename": "refseq_protein_all_10090.gpff",
                    "uri": "https:/data.glygen.org/ln2downloads/ncbi/refseq/2023_10_12/refseq_protein_all_10090.gpff"
                }
            }
        ],
        "output_subdomain": [
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.3.1/reviewed/mouse_protein_info_refseq.csv",
                    "filename": "mouse_protein_info_refseq.csv"
                }
            },
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.3.1/reviewed/mouse_protein_info_refseq.stat.csv",
                    "filename": "mouse_protein_info_refseq.stat.csv"
                }
            }
        ]
    }
}