{
    "object_id": "https://biocomputeobject.org/GLY_000134/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": "Mus musculus",
                        "category_name": "species"
                    },
                    {
                        "category_value": "Protein",
                        "category_name": "molecule"
                    },
                    {
                        "category_value": "Protein Function",
                        "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": "Mouse Protein Function (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": "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": "2019-03-16T15:38:06.344586",
        "modified": "2022-09-26T08:56:48.530675",
        "version": "v-2.0.3"
    },
    "description_domain": {
        "keywords": [
            "protein function",
            "NCBI RefSeq",
            "geneRIF",
            "PMID"
        ],
        "pipeline_steps": [
            {
                "name": "extract-from-gp.py",
                "step_number": 1,
                "version": "1.0",
                "input_list": [
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-2.0.3/source/refseq_protein_all_10090.gpff",
                        "filename": "refseq_protein_all_10090.gpff"
                    },
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-2.0.3/reviewed/mouse_protein_xref_refseq.csv",
                        "filename": "mouse_protein_xref_refseq.csv"
                    }
                ],
                "output_list": [
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-2.0.3/reviewed/mouse_protein_function_refseq.csv",
                        "filename": "mouse_protein_function_refseq.csv"
                    }
                ],
                "prerequisite": [],
                "description": "The script extract-from-gp.py extracts protein function information from NCBI RefSeq gpff format files and maps it to the UniProtKb canonical accessions for a final dataset output in a csv format. "
            }
        ]
    },
    "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/extract-from-endpoint.py",
                    "filename": "extract-from-endpoint.py"
                }
            }
        ]
    },
    "error_domain": {
        "empirical_error": {},
        "algorithmic_error": {}
    },
    "parametric_domain": [
        {
            "step": "",
            "param": "",
            "value": ""
        }
    ],
    "usability_domain": [
        "The Mouse Protein Function (NCBI RefSeq) dataset contains biological function annotation of mouse [taxid:10090] genes from GeneRIF section in the 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": "e3577fdaa20811edab60e54b70f2df03",
    "io_domain": {
        "input_subdomain": [
            {
                "uri": {
                    "filename": "mouse_protein_xref_refseq.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.0.3/reviewed/mouse_protein_xref_refseq.csv"
                }
            },
            {
                "uri": {
                    "filename": "refseq_protein_all_10090.gpff",
                    "uri": "/data/shared/glyds/downloads/ncbi/refseq/2022_10/refseq_protein_all_10090.gpff"
                }
            }
        ],
        "output_subdomain": [
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.0.3/reviewed/mouse_protein_function_refseq.csv",
                    "filename": "mouse_protein_function_refseq.csv"
                }
            },
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.0.3/reviewed/mouse_protein_function_refseq.stat.csv",
                    "filename": "mouse_protein_function_refseq.stat.csv"
                }
            }
        ]
    }
}