{
    "object_id": "https://biocomputeobject.org/GLY_000249/1.8.25", 
    "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": "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": "Rat 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"
            }
        ], 
        "created": "2019-09-06T14:06:38.975527", 
        "modified": "2020-04-10T11:59:41.589685", 
        "version": "1.8.25"
    }, 
    "description_domain": {
        "keywords": [
            "protein function", 
            "NCBI RefSeq", 
            "geneRIF", 
            "PMID"
        ], 
        "pipeline_steps": [
            {
                "name": "make-protein-dataset.py", 
                "step_number": 1, 
                "version": "", 
                "input_list": [
                    {
                        "uri": "https://data.glygen.org/ln2wwwdata/reviewed/rat_protein_xref_refseq.csv", 
                        "filename": "rat_protein_xref_refseq.csv"
                    }, 
                    {
                        "uri": "https://data.glygen.org/ln2wwwdata/source/refseq_protein_all_10116.gpff", 
                        "filename": "refseq_protein_all_10116.gpff"
                    }
                ], 
                "output_list": [
                    {
                        "uri": "https://data.glygen.org/ln2wwwdata/reviewed/rat_protein_info_refseq.csv", 
                        "filename": "rat_protein_info_refseq.csv"
                    }
                ], 
                "prerequisite": [], 
                "description": "The script parses the NCBI RefSeq gpff files to extract the geneRIFs that contain protein function. The script also maps 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/GW-HIVE/glygen-backend-integration/blob/master/pipeline/integrator/make-protein-dataset.py", 
                    "filename": "make-protein-dataset.py"
                }
            }
        ]
    }, 
    "usability_domain": [
        "The Rat Protein Function (NCBI RefSeq) dataset contains biological function annotaion of rat [taxid:10116] 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": "cc9f3df76057996a785870ba6a13e244", 
    "_id": {
        "$oid": "5d72a02ef5c47b0f4342336a"
    }, 
    "io_domain": {
        "input_subdomain": [
            {
                "uri": {
                    "uri": "https://data.glygen.org//ln2downloads/ncbi/refseq/2020_12/refseq_protein_all_10116.gpff", 
                    "filename": "refseq_protein_all_10116.gpff"
                }
            }, 
            {
                "uri": {
                    "uri": "https://data.glygen.org//ln2releases/data/v-1.8.25/reviewed/rat_protein_xref_refseq.csv", 
                    "filename": "rat_protein_xref_refseq.csv"
                }
            }
        ], 
        "output_subdomain": [
            {
                "mediatype": "csv", 
                "uri": {
                    "uri": "http://data.glygen.org/ln2wwwdata/reviewed/rat_protein_function_refseq.csv", 
                    "filename": "rat_protein_function_refseq.csv"
                }
            }, 
            {
                "mediatype": "csv", 
                "uri": {
                    "uri": "http://data.glygen.org/ln2wwwdata/reviewed/rat_protein_function_refseq.stat.csv", 
                    "filename": "rat_protein_function_refseq.stat.csv"
                }
            }
        ]
    }
}