{
    "io_domain": {
        "input_subdomain": [
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2wwwdata/reviewed/rat_proteoform_glycosylation_sites_glyconnect.csv", 
                    "filename": "rat_proteoform_glycosylation_sites_glyconnect.csv"
                }
            }
        ], 
        "output_subdomain": [
            {
                "mediatype": "csv", 
                "uri": {
                    "uri": "http://data.glygen.org/ln2wwwdata/reviewed/rat_proteoform_citations_glyconnect.csv", 
                    "filename": "rat_proteoform_citations_glyconnect.csv"
                }
            }, 
            {
                "mediatype": "csv", 
                "uri": {
                    "uri": "http://data.glygen.org/ln2wwwdata/reviewed/rat_proteoform_citations_glyconnect.stat.csv", 
                    "filename": "rat_proteoform_citations_glyconnect.stat.csv"
                }
            }
        ]
    }, 
    "spec_version": "1.3.0", 
    "object_id": "https://biocomputeobject.org/GLY_000485/1.7.10", 
    "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": "Proteoform", 
                        "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": "Rat Proteoform Citations (GlyConnect)", 
        "contributors": [
            {
                "affiliation": "The George Washington University", 
                "contribution": [
                    "createdBy"
                ], 
                "name": "Rahi Navelkar", 
                "email": "rsn13@gwu.edu"
            }, 
            {
                "affiliation": "The George Washington University", 
                "contribution": [
                    "createdBy"
                ], 
                "name": "Jeet Vora", 
                "email": "jeetvora@gwu.edu"
            }, 
            {
                "affiliation": "The George Washington University", 
                "contribution": [
                    "createdBy"
                ], 
                "name": "Robel Kahsay", 
                "email": "rykahsay@gwu.edu"
            }, 
            {
                "affiliation": "GlyConnect", 
                "contribution": [
                    "contributedBy"
                ], 
                "name": "Frederique Lisacek", 
                "email": "frederique.lisacek@sib.swiss"
            }, 
            {
                "affiliation": "GlyConnect", 
                "contribution": [
                    "contributedBy"
                ], 
                "name": "Julien Mariethoz", 
                "email": "julien.mariethoz@sib.swiss"
            }
        ], 
        "created": "2020-04-28T11:07:38.533264", 
        "modified": "2020-04-28T11:09:38.064805", 
        "version": "1.7.10"
    }, 
    "description_domain": {
        "keywords": [
            "PMID", 
            "journal", 
            "citations", 
            "authors", 
            "GlyConnect", 
            "PubMed"
        ], 
        "pipeline_steps": [
            {
                "name": "make-proteoform-dataset.py", 
                "step_number": 1, 
                "version": "", 
                "input_list": [
                    {
                        "uri": "https://data.glygen.org/ln2wwwdata/reviewed/rat_proteoform_glycosylation_sites_glyconnect.csv", 
                        "filename": "rat_proteoform_glycosylation_sites_glyconnect.csv"
                    }
                ], 
                "output_list": [
                    {
                        "uri": "https://data.glygen.org/ln2wwwdata/reviewed/rat_proteoform_citations_glyconnect.csv", 
                        "filename": "rat_proteoform_citations_glyconnect.csv"
                    }
                ], 
                "prerequisite": [], 
                "description": "The script parses the input file for PMIDs and accesses the NCBI Entrez infrastructure to download the publication information of the corresponding PMIDs"
            }
        ]
    }, 
    "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-proteoform-dataset.py", 
                    "filename": "make-proteoform-dataset.py"
                }
            }
        ]
    }, 
    "parametric_domain": [
        {
            "step": "", 
            "value": "", 
            "param": ""
        }
    ], 
    "usability_domain": [
        "The Rat Proteoform Citations (GlyConnect) dataset provides citation details corresponding to the PubMed ID (PMID) from file https://data.glygen.org/GLYDS000329. The script accesses the NCBI ENTREZ infrastructure to retrieve the publication information of the given PMID. The dataset is contributed by GlyGen and processed by GlyGen. If you use this dataset please provide proper attribution to GlyConnect (https://glyconnect.expasy.org/)."
    ], 
    "etag": "403b5b53ba24b1648f6e6ce76c17e48a", 
    "_id": {
        "$oid": "5ea846baf5c47b213b416739"
    }
}