{
    "io_domain": {
        "input_subdomain": [
            {
                "uri": {
                    "access_time": "", 
                    "sha1_chksum": "", 
                    "uri": "https://data.glygen.org//ln2releases/data/v-1.9.9/reviewed/sarscov2_proteoform_glycosylation_sites_glyconnect.csv", 
                    "filename": "sarscov2_proteoform_glycosylation_sites_glyconnect.csv"
                }
            }
        ], 
        "output_subdomain": [
            {
                "mediatype": "csv", 
                "uri": {
                    "access_time": "2020-12-11T14:12:34.190176", 
                    "sha1_chksum": "", 
                    "uri": "http://data.glygen.org/ln2wwwdata/reviewed/sarscov2_proteoform_citations_glycosylation_glyconnect.csv", 
                    "filename": "sarscov2_proteoform_citations_glycosylation_glyconnect.csv"
                }
            }, 
            {
                "mediatype": "csv", 
                "uri": {
                    "access_time": "", 
                    "sha1_chksum": "", 
                    "uri": "http://data.glygen.org/ln2wwwdata/reviewed/sarscov2_proteoform_citations_glyconnect.stat.csv", 
                    "filename": "sarscov2_proteoform_citations_glyconnect.stat.csv"
                }
            }
        ]
    }, 
    "bco_spec_version": "1.3", 
    "checksum": "d1325d974426bed19e57b55c46a395a0", 
    "extension_domain": {
        "additional_license": {
            "data_license": "https://creativecommons.org/licenses/by/4.0/", 
            "script_license": "https://www.gnu.org/licenses/gpl-3.0.en.html"
        }, 
        "fhir_extension": [], 
        "scm_extension": {
            "scm_repository": "", 
            "scm_path": "", 
            "scm_type": "git", 
            "scm_commit": ""
        }, 
        "dataset_categories": [
            {
                "category_value": "SARS coronavirus (SARS-CoV-2 or 2019-nCoV)", 
                "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": "", 
        "name": "SARS-CoV2 Proteoform Citations (GlyConnect)", 
        "contributors": [
            {
                "orcid": "", 
                "affiliation": "The George Washington University", 
                "contribution": [
                    "createdBy"
                ], 
                "name": "Rahi Navelkar", 
                "email": "rsn13@gwu.edu"
            }, 
            {
                "orcid": "", 
                "affiliation": "The George Washington University", 
                "contribution": [
                    "createdBy"
                ], 
                "name": "Jeet Vora", 
                "email": "jeetvora@gwu.edu"
            }, 
            {
                "orcid": "", 
                "affiliation": "The George Washington University", 
                "contribution": [
                    "createdBy"
                ], 
                "name": "Robel Kahsay", 
                "email": "rykahsay@gwu.edu"
            }, 
            {
                "orcid": "", 
                "affiliation": "GlyConnect", 
                "contribution": [
                    "contributedBy"
                ], 
                "name": "Frederique Lisacek", 
                "email": "frederique.lisacek@sib.swiss"
            }, 
            {
                "orcid": "", 
                "affiliation": "GlyConnect", 
                "contribution": [
                    "contributedBy"
                ], 
                "name": "Julien Mariethoz", 
                "email": "julien.mariethoz@sib.swiss"
            }
        ], 
        "created": "2020-12-11T14:19:01.972822", 
        "modified": "2020-12-11T14:24:00.759472", 
        "version": "1.9.9"
    }, 
    "description_domain": {
        "keywords": [
            "PMID", 
            "journal", 
            "citations", 
            "authors", 
            "GlyConnect", 
            "PubMed"
        ], 
        "pipeline_steps": [
            {
                "name": "make-proteoform-dataset.py", 
                "step_number": 1, 
                "version": "", 
                "input_list": [
                    {
                        "access_time": "2020-12-11T14:10:34.190176", 
                        "uri": "https://data.glygen.org/ln2wwwdata/reviewed/sarscov2_proteoform_glycosylation_sites_glyconnect.csv", 
                        "filename": "sarscov2_proteoform_glycosylation_sites_glyconnect.csv"
                    }
                ], 
                "output_list": [
                    {
                        "access_time": "2020-12-11T14:10:34.190176", 
                        "uri": "https://data.glygen.org/ln2wwwdata/reviewed/sarscov2_proteoform_citations_glyconnect.csv", 
                        "filename": "sarscov2_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": {
                    "access_time": "2020-12-11T14:10:34.190176", 
                    "sha1_chksum": "", 
                    "uri": "https://www.python.org/download/releases/2.7.5/"
                }
            }
        ], 
        "script": [
            {
                "uri": {
                    "access_time": "2020-12-11T14:10:34.190176", 
                    "sha1_chksum": "", 
                    "uri": "https://github.com/GW-HIVE/glygen-backend-integration/blob/master/pipeline/integrator/make-proteoform-dataset.py", 
                    "filename": "make-proteoform-dataset.py"
                }
            }
        ]
    }, 
    "error_domain": {
        "empirical_error": {}, 
        "algorithmic_error": {}
    }, 
    "parametric_domain": [
        {
            "step": "", 
            "value": "", 
            "param": ""
        }
    ], 
    "usability_domain": [
        "The SARS-CoV2 Proteoform Citations (GlyConnect) dataset provides citation details corresponding to the PubMed ID (PMID) from file https://data.glygen.org/GLYDS000530. 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/)."
    ], 
    "_id": {
        "$oid": "5fd3c625f5c47b37985ec301"
    }, 
    "bco_id": "http://data.glygen.org/GLY_000535"
}