{
    "io_domain": {
        "input_subdomain": [
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2data/releases/data/v-1.0.13/source/refseq_protein_all_9606.gpff",
                    "filename": "refseq_protein_all_9606.gpff"
                }
            },
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2data/releases/data/v-1.0.13/reviewed/human_protein_idmapping.csv",
                    "filename": "human_protein_idmapping.csv"
                }
            }
        ],
        "output_subdomain": [
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https://data.glygen.org/ln2data/releases/data/v-1.0.13/reviewed/human_protein_function_refseq.csv",
                    "filename": "human_protein_function_refseq.csv"
                }
            },
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https://data.glygen.org/ln2data/releases/data/v-1.0.13/reviewed/human_protein_function_refseq.stat.csv",
                    "filename": "human_protein_function_refseq.stat.csv"
                }
            }
        ]
    },
    "spec_version": "1.3.0",
    "object_id": "https://biocomputeobject.org/GLY_000133/v-1.0.13",
    "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"
                    }
                ]
            }
        }
    ],
    "provenance_domain": {
        "license": "https://creativecommons.org/licenses/by/4.0/",
        "name": "Human Refseq Protein Function",
        "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"
            }
        ],
        "created": "2019-03-16T15:35:15.248632",
        "modified": "2019-04-24T10:47:03.644563",
        "version": "v-1.0.13"
    },
    "description_domain": {
        "keywords": [
            "protein function",
            "NCBI RefSeq"
        ],
        "pipeline_steps": [
            {
                "name": "extract-from-gp.py",
                "step_number": 1,
                "version": "1.0",
                "input_list": [
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-1.0.13/source/refseq_protein_all_9606.gpff",
                        "filename": "refseq_protein_all_9606.gpff"
                    },
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-1.0.13/reviewed/human_protein_idmapping.csv",
                        "filename": "human_protein_idmapping.csv"
                    }
                ],
                "output_list": [
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-1.0.13/reviewed/human_protein_function_refseq.csv",
                        "filename": "human_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/",
                    "filename": ""
                }
            }
        ],
        "script": [
            {
                "uri": {
                    "uri": "https://github.com/glygener/glygen-backend-integration/blob/master/pipeline/extractor/extract-from-gp.py",
                    "filename": "extract-from-gp.py"
                }
            }
        ]
    },
    "error_domain": {
        "empirical_error": {},
        "algorithmic_error": {},
        "emperical_error": {}
    },
    "parametric_domain": [
        {
            "step": "",
            "value": "",
            "param": ""
        }
    ],
    "usability_domain": [
        "The Human RefSeq Protein Function dataset contains biological function annotaion of human [taxid:9606] genes from GeneRIF section in the NCBI RefSeq database",
        "The data is downloaded from NCBI RefSeq and is processed into a dataset by GlyGen. If you use this dataset please provide proper attribution to NCBI RefSeq and GlyGen"
    ],
    "etag": "1dbb2d98beaba72bc9a1682a73d0b66f",
    "_id": {
        "$oid": "5c8d4ff3f5c47b31a13ff81c"
    }
}
