{
    "io_domain": {
        "input_subdomain": [
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2wwwdata/source/refseq_protein_all_viral.gpff", 
                    "filename": "refseq_protein_all_viral.gpff"
                }
            }, 
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2wwwdata/reviewed/sarscov1_protein_xref_refseq.csv", 
                    "filename": "sarscov1_protein_xref_refseq.csv"
                }
            }
        ], 
        "output_subdomain": [
            {
                "mediatype": "csv", 
                "uri": {
                    "uri": "http://data.glygen.org/ln2wwwdata/reviewed/sarscov1_protein_citations_refseq.csv", 
                    "filename": "sarscov1_protein_citations_refseq.csv"
                }
            }, 
            {
                "mediatype": "csv", 
                "uri": {
                    "uri": "http://data.glygen.org/ln2wwwdata/reviewed/sarscov1_protein_citations_refseq.stat.csv", 
                    "filename": "sarscov1_protein_citations_refseq.stat.csv"
                }
            }
        ]
    }, 
    "spec_version": "1.3.0", 
    "object_id": "https://biocomputeobject.org/GLY_000417/1.5.30", 
    "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-1)", 
                        "category_name": "species"
                    }, 
                    {
                        "category_value": "Protein", 
                        "category_name": "molecule"
                    }, 
                    {
                        "category_value": "Protein Citations", 
                        "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-CoV1 Protein Citations (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"
            }
        ], 
        "created": "2020-03-15T13:32:44.009025", 
        "modified": "2020-04-09T18:39:36.624800", 
        "version": "1.5.30"
    }, 
    "description_domain": {
        "keywords": [
            "PMID", 
            "journal", 
            "citations", 
            "authors", 
            "RefSeq", 
            "PubMed"
        ], 
        "pipeline_steps": [
            {
                "name": "make-protein-dataset.py", 
                "step_number": 1, 
                "version": "", 
                "input_list": [
                    {
                        "uri": "https://data.glygen.org/ln2wwwdata/reviewed/sarscov1_protein_xref_refseq.csv", 
                        "filename": "sarscov1_protein_xref_refseq.csv"
                    }, 
                    {
                        "uri": "https://data.glygen.org/ln2wwwdata/source/refseq_protein_all_viral.gpff", 
                        "filename": "refseq_protein_all_viral.gpff"
                    }
                ], 
                "output_list": [
                    {
                        "uri": "https://data.glygen.org/ln2wwwdata/reviewed/sarscov1_protein_citations_refseq.csv", 
                        "filename": "sarscov1_protein_citations_refseq.csv"
                    }
                ], 
                "prerequisite": [], 
                "description": "The script parses the NCBI RefSeq gpff files to extract the publication information. 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 SARS coronavirus (SARS-CoV-1) Protein Citations (RefSeq) datasets contains sarscov1 [taxid:694009] UniProtKB accessions mapped to publication information (pmid, title, journal name, publication, date, authors of corresponding RefSeq accessions. 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": "047710f45f760a76631a34e1d872fd3d", 
    "_id": {
        "$oid": "5e86a7fcf5c47b1a53100d9e"
    }
}
