{
    "io_domain": {
        "input_subdomain": [
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2wwwdata/source/uniprot-proteome-sars-cov-2.nt", 
                    "filename": "uniprot-proteome-sars-cov-2.nt"
                }
            }, 
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2wwwdata/reviewed/sarscov2_protein_masterlist.csv", 
                    "filename": "sarscov2_protein_masterlist.csv"
                }
            }
        ], 
        "output_subdomain": [
            {
                "mediatype": "csv", 
                "uri": {
                    "uri": "http://data.glygen.org/ln2wwwdata/reviewed/sarscov2_protein_go_annotation.csv", 
                    "filename": "sarscov2_protein_go_annotation.csv"
                }
            }, 
            {
                "mediatype": "csv", 
                "uri": {
                    "uri": "http://data.glygen.org/ln2wwwdata/reviewed/sarscov2_protein_go_annotation.stat.csv", 
                    "filename": "sarscov2_protein_go_annotation.stat.csv"
                }
            }
        ]
    }, 
    "spec_version": "1.3.0", 
    "object_id": "https://biocomputeobject.org/GLY_000430/1.5.30", 
    "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": "SARS coronavirus (SARS-CoV-2 or 2019-nCoV)", 
                        "category_name": "species"
                    }, 
                    {
                        "category_value": "Protein", 
                        "category_name": "molecule"
                    }, 
                    {
                        "category_value": "Gene Ontology (Go)", 
                        "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": "SARS-CoV2 Gene Ontology (GO)", 
        "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": "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"
            }
        ], 
        "created": "2020-04-03T11:46:41.584409", 
        "modified": "2020-04-10T12:26:18.336828", 
        "version": "1.5.30"
    }, 
    "description_domain": {
        "keywords": [
            "go id", 
            "go name", 
            "molecular function", 
            "biological process", 
            "cellular component", 
            "Gene Ontology (GO)"
        ], 
        "pipeline_steps": [
            {
                "name": "make-protein-dataset.py", 
                "step_number": 1, 
                "version": "", 
                "input_list": [
                    {
                        "uri": "https://data.glygen.org/ln2wwwdata/reviewed/sarscov2_protein_masterlist.csv", 
                        "filename": "sarscov2_protein_masterlist.csv"
                    }, 
                    {
                        "uri": "https://data.glygen.org/ln2wwwdata/source/uniprot-proteome-sars-cov-2.nt", 
                        "filename": "uniprot-proteome-sars-cov-2.nt"
                    }
                ], 
                "output_list": [
                    {
                        "uri": "https://data.glygen.org/ln2wwwdata/reviewed/sarscov2_protein_go_annotation.csv", 
                        "filename": "sarscov2_protein_go_annotation.csv"
                    }
                ], 
                "prerequisite": [], 
                "description": "The script extracts the GO Id, name, category and evidence from the triplestore of NT file and maps it to the UniProtKB canonical sequnces"
            }
        ]
    }, 
    "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 SARS coronavirus (SARS-CoV-2 or 2019-nCov) Gene Ontology (GO) dataset contains sarscov2 [taxid:2697049] Gene Ontology Id and name, categories and evidence information for UniProtKB canonical accessions. The dataset is derived from 2020-01 UniProt release. If you use this dataset please provide proper attribution to EMBL-EBI-UniProtKB and GlyGen"
    ], 
    "etag": "fbf7a948d6d1821ea99a979f9566bfa8", 
    "_id": {
        "$oid": "5e8749f8f5c47b6760977032"
    }
}
