{
    "io_domain": {
        "input_subdomain": [
            {
                "uri": {
                    "access_time": "", 
                    "sha1_chksum": "", 
                    "uri": "https://data.glygen.org//ln2releases/data/v-1.11.2/reviewed/hcv1a_protein_masterlist.csv", 
                    "filename": "hcv1a_protein_masterlist.csv"
                }
            }, 
            {
                "uri": {
                    "access_time": "", 
                    "sha1_chksum": "", 
                    "uri": "https://data.glygen.org//ln2downloads/ebi/2021_03/uniprot-proteome-hepatitis-c-virus-1a.nt", 
                    "filename": "uniprot-proteome-hepatitis-c-virus-1a.nt"
                }
            }
        ], 
        "output_subdomain": [
            {
                "mediatype": "csv", 
                "uri": {
                    "access_time": "", 
                    "sha1_chksum": "", 
                    "uri": "http://data.glygen.org/datasets/reviewed/hcv1a_proteoform_glycosylation_sites_uniprotkb.csv", 
                    "filename": "hcv1a_proteoform_glycosylation_sites_uniprotkb.csv"
                }
            }, 
            {
                "mediatype": "csv", 
                "uri": {
                    "access_time": "", 
                    "sha1_chksum": "", 
                    "uri": "http://data.glygen.org/datasets/reviewed/hcv1a_proteoform_glycosylation_sites_uniprotkb.stat.csv", 
                    "filename": "hcv1a_proteoform_glycosylation_sites_uniprotkb.stat.csv"
                }
            }
        ]
    }, 
    "bco_spec_version": "1.3.0", 
    "checksum": "1ccf09f6515c9db345019e5f2380c7b7", 
    "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": "Hepatitis C virus (genotype 1a, isolate H)", 
                "category_name": "species"
            }, 
            {
                "category_value": "Proteoform", 
                "category_name": "molecule"
            }, 
            {
                "category_value": "Glycosylation Sites", 
                "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": "HCV1a Glycosylation Sites (UniProtKB)", 
        "contributors": [
            {
                "orcid": "", 
                "affiliation": "The George Washington University", 
                "contribution": [
                    "createdBy"
                ], 
                "name": "Jeet Vora", 
                "email": "jeetvora@gwu.edu"
            }, 
            {
                "orcid": "", 
                "affiliation": "The George Washington University", 
                "contribution": [
                    "createdBy"
                ], 
                "name": "Rahi Navelkar", 
                "email": "rsn13@gwu.edu"
            }, 
            {
                "orcid": "", 
                "affiliation": "The George Washington University", 
                "contribution": [
                    "createdBy"
                ], 
                "name": "Robel Kahsay", 
                "email": "rykahsay@gwu.edu"
            }, 
            {
                "orcid": "", 
                "affiliation": "EMBL-EBI-UniProt", 
                "contribution": [
                    "contributedBy"
                ], 
                "name": "Maria Martin", 
                "email": "martin@ebi.ac.uk"
            }, 
            {
                "orcid": "", 
                "affiliation": "EMBL-EBI-UniProt", 
                "contribution": [
                    "contributedBy"
                ], 
                "name": "Preethi Vasudev", 
                "email": "pvasudev@ebi.ac.uk"
            }, 
            {
                "orcid": "", 
                "affiliation": "The George Washington University", 
                "contribution": [
                    "createdBy"
                ], 
                "name": "Karina Martinez", 
                "email": "karinamartinez@gwu.edu"
            }, 
            {
                "orcid": "", 
                "affiliation": "The George Washington University", 
                "contribution": [
                    "createdBy"
                ], 
                "name": "Amanda Bell", 
                "email": "amandab2140@gwu.edu"
            }, 
            {
                "orcid": "", 
                "affiliation": "The George Washington University", 
                "contribution": [
                    "createdBy"
                ], 
                "name": "Jonathon Keeney", 
                "email": "keeneyjg@gwu.edu"
            }, 
            {
                "orcid": "", 
                "affiliation": "The George Washington University", 
                "contribution": [
                    "createdBy"
                ], 
                "name": "Millicent Quartey", 
                "email": "mquartey@email.gwu.edu"
            }
        ], 
        "created": "2020-03-15T12:24:50.340202", 
        "modified": "2020-04-09T17:54:08.766039", 
        "version": "1.11.2"
    }, 
    "description_domain": {
        "keywords": [
            "glycosylation", 
            "reported glycosylation", 
            "predicted glycosylation", 
            "UniProtKB", 
            "PubMed", 
            "Evidence and conclusion Ontology"
        ], 
        "pipeline_steps": [
            {
                "name": "make-proteoform-dataset.py", 
                "step_number": 1, 
                "version": "", 
                "input_list": [
                    {
                        "access_time": "", 
                        "uri": "https://data.glygen.org/ln2wwwdata/reviewed/hcv1a_protein_masterlist.csv", 
                        "filename": "hcv1a_protein_masterlist.csv"
                    }, 
                    {
                        "access_time": "", 
                        "uri": "https://data.glygen.org/ln2wwwdata/source/uniprot-proteome-hepatitis-c-virus-1a.nt", 
                        "filename": "uniprot-proteome-hepatitis-c-virus-1a.nt"
                    }
                ], 
                "output_list": [
                    {
                        "access_time": "", 
                        "uri": "https://data.glygen.org/ln2wwwdata/reviewed/hcv1a_proteoform_glycosylation_sites_uniprotkb.csv", 
                        "filename": "hcv1a_proteoform_glycosylation_sites_uniprotkb.csv"
                    }
                ], 
                "prerequisite": [], 
                "description": "The script loads the glycosylation information from the triplestore conatining triples from the nt file and extracts as a csv dataset"
            }
        ]
    }, 
    "execution_domain": {
        "external_data_endpoints": [], 
        "environment_variables": {}, 
        "script_driver": "manual", 
        "software_prerequisites": [
            {
                "version": "2.7.5", 
                "name": "Python", 
                "uri": {
                    "access_time": "2019-01-07T17:43:17-0500", 
                    "sha1_chksum": "", 
                    "uri": "https://www.python.org/download/releases/2.7.5/"
                }
            }
        ], 
        "script": [
            {
                "uri": {
                    "access_time": "", 
                    "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": {}, 
        "emperical_error": {}
    }, 
    "parametric_domain": [
        {
            "step": "", 
            "value": "", 
            "param": ""
        }
    ], 
    "usability_domain": [
        "The Hepatitis C virus (genotype 1a, isolate H) Glycosylation Sites (UniProtKB) dataset contains hcv1a [taxid:11108] UniProtKB canonical accessions with information on reported and predicted glycosylation sites in UniProtKB database. The dataset is derived from 2019-09 UniProtKB release. If you use this dataset please provide proper attribution to EMBL-EBI-UniProtKB and GlyGen"
    ], 
    "_id": {
        "$oid": "5e6e56d2f5c47b5905c7e3c0"
    }, 
    "bco_id": "http://data.glygen.org/GLY_000382"
}