{
    "io_domain": {
        "input_subdomain": [
            {
                "uri": {
                    "access_time": "", 
                    "sha1_chksum": "", 
                    "uri": "https://data.glygen.org//ln2downloads/glycosmos/04_08_2021/ggdonto.ttl", 
                    "filename": "ggdonto.ttl"
                }
            }, 
            {
                "uri": {
                    "access_time": "", 
                    "sha1_chksum": "", 
                    "uri": "https://data.glygen.org//ln2releases/data/v-1.11.2/reviewed/human_protein_xref_geneid.csv", 
                    "filename": "human_protein_xref_geneid.csv"
                }
            }, 
            {
                "uri": {
                    "access_time": "", 
                    "sha1_chksum": "", 
                    "uri": "https://data.glygen.org//ln2releases/data/v-1.11.2/reviewed/protein_disease_idmap.csv", 
                    "filename": "protein_disease_idmap.csv"
                }
            }
        ], 
        "output_subdomain": [
            {
                "mediatype": "csv", 
                "uri": {
                    "access_time": "2020-09-24T13:36:03.947890", 
                    "sha1_chksum": "", 
                    "uri": "http://data.glygen.org/ln2wwwdata/reviewed/human_protein_disease_glycosmos.csv", 
                    "filename": "human_protein_disease_glycosmos.csv"
                }
            }, 
            {
                "mediatype": "csv", 
                "uri": {
                    "access_time": "", 
                    "sha1_chksum": "", 
                    "uri": "http://data.glygen.org/ln2wwwdata/reviewed/human_protein_disease_glycosmos.stat.csv", 
                    "filename": "human_protein_disease_glycosmos.stat.csv"
                }
            }
        ]
    }, 
    "bco_spec_version": "1.3", 
    "checksum": "2bec35753f9b6a5be7e8ce07dd630021", 
    "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": "Homo sapiens", 
                "category_name": "species"
            }, 
            {
                "category_value": "Protein", 
                "category_name": "molecule"
            }, 
            {
                "category_value": "Disease GlyCosmos", 
                "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": "Human Protein Diseases (GlyCosmos)", 
        "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": "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"
            }, 
            {
                "orcid": "", 
                "affiliation": "", 
                "contribution": [
                    "contributedBy"
                ], 
                "name": "Not Applicable", 
                "email": ""
            }
        ], 
        "created": "2020-09-24T13:34:03.947890", 
        "modified": "2022-01-18T12:24:09.466735", 
        "version": "1.11.2"
    }, 
    "description_domain": {
        "keywords": [
            "Gene", 
            "Protein", 
            "Disease", 
            "GlyCosmos"
        ], 
        "pipeline_steps": [
            {
                "name": "", 
                "step_number": 0, 
                "version": "", 
                "input_list": [
                    {
                        "access_time": "2020-09-24T13:34:03.947890", 
                        "sha1_chksum": "", 
                        "uri": "http://data.glygen.org/ln2wwwdata/reviewed/human_protein_xref_geneid.csv", 
                        "filename": "human_protein_xref_geneid.csv"
                    }, 
                    {
                        "access_time": "2020-09-24T13:34:03.947890", 
                        "sha1_chksum": "", 
                        "uri": "http://data.glygen.org/ln2wwwdata/source/ggdonto.ttl", 
                        "filename": "ggdonto.ttl"
                    }, 
                    {
                        "access_time": "2020-09-24T13:34:03.947890", 
                        "sha1_chksum": "", 
                        "uri": "http://data.glygen.org/ln2wwwdata/source/protein_disease_idmap.csv", 
                        "filename": "protein_disease_idmap.csv"
                    }
                ], 
                "output_list": [
                    {
                        "access_time": "2020-09-24T13:36:03.947890", 
                        "sha1_chksum": "", 
                        "uri": "http://data.glygen.org/ln2wwwdata/reviewed/ human_protein_disease_glycosmos.csv", 
                        "filename": " human_protein_disease_glycosmos.csv"
                    }
                ], 
                "prerequisite": [], 
                "description": "The file is downloaded from the GlyCosmos database directly. The following steps are done to generate the final output file: 1. Mapping gene_id to UniProtKB accession from input file 1 2) Map the dd_uri to mim_id and disease name from input file 2 3) Map mim_id to doid and mongo ids using input file 3. "
            }
        ]
    }, 
    "execution_domain": {
        "external_data_endpoints": [], 
        "environment_variables": {}, 
        "script_driver": "manual", 
        "software_prerequisites": [
            {
                "version": "2.7.5", 
                "name": "Python", 
                "uri": {
                    "access_time": "2020-09-24T13:34:03.947890", 
                    "sha1_chksum": "", 
                    "uri": "https://github.com/GW-HIVE/glygen-backend-integration/blob/master/pipeline/integrator/", 
                    "filename": ""
                }
            }
        ], 
        "script": [
            {
                "uri": {
                    "access_time": "", 
                    "sha1_chksum": "", 
                    "uri": "", 
                    "filename": ""
                }
            }
        ]
    }, 
    "error_domain": {
        "empirical_error": {}, 
        "algorithmic_error": {}
    }, 
    "parametric_domain": [
        {
            "step": "", 
            "value": "", 
            "param": ""
        }
    ], 
    "usability_domain": [
        "The file contain disease data associated to protein/gene downloaded from GlyCosmos database. [https://glycosmos.org/homes/download]. "
    ], 
    "_id": {
        "$oid": "5f6cd88bf5c47b0de0793525"
    }, 
    "bco_id": "http://data.glygen.org/GLY_000529"
}