{
    "io_domain": {
        "input_subdomain": [
            {
                "uri": {
                    "access_time": "", 
                    "sha1_chksum": "", 
                    "uri": "https://data.glygen.org//ln2releases/data/v-1.11.2/reviewed/human_protein_masterlist.csv", 
                    "filename": "human_protein_masterlist.csv"
                }
            }, 
            {
                "uri": {
                    "access_time": "", 
                    "sha1_chksum": "", 
                    "uri": "https://data.glygen.org//ln2releases/data/v-1.11.2/reviewed/human_protein_xref_pharos.csv", 
                    "filename": "human_protein_xref_pharos.csv"
                }
            }
        ], 
        "output_subdomain": [
            {
                "mediatype": "csv", 
                "uri": {
                    "access_time": "", 
                    "sha1_chksum": "", 
                    "uri": "https://data.glygen.org/ln2wwwdata/reviewed/protein_glygen_pharos_xref_mapping.csv", 
                    "filename": "protein_glygen_pharos_xref_mapping.csv"
                }
            }, 
            {
                "mediatype": "txt", 
                "uri": {
                    "access_time": "", 
                    "sha1_chksum": "", 
                    "uri": "https://data.glygen.org/ln2wwwdata/logs/glygen_pharos_xref_mapping.log", 
                    "filename": "glygen_pharos_xref_mapping.log"
                }
            }, 
            {
                "mediatype": "csv", 
                "uri": {
                    "access_time": "", 
                    "sha1_chksum": "", 
                    "uri": "https://data.glygen.org/ln2wwwdata/reviewed/protein_glygen_pharos_xref_mapping.stat.csv", 
                    "filename": "protein_glygen_pharos_xref_mapping.stat.csv"
                }
            }
        ]
    }, 
    "bco_spec_version": "1.3.0", 
    "checksum": "83e3c8e5ce8866ae34e9429b68c89bd2", 
    "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": "GlyGen-Pharos Crossreferences", 
                "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": "GlyGen-Pharos Crossreferences Mapping", 
        "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": "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": "2020-04-03T22:31:02.233246", 
        "modified": "2022-01-13T16:35:01.358361", 
        "version": "1.11.2"
    }, 
    "description_domain": {
        "keywords": [
            "Pharos", 
            "GlyGen", 
            "crossreferences", 
            "human", 
            "mouse", 
            "rat"
        ], 
        "pipeline_steps": [
            {
                "name": "make-species-agnostic-dataset.py", 
                "step_number": 1, 
                "version": "", 
                "input_list": [
                    {
                        "access_time": "", 
                        "uri": "https://data.glygen.org/ln2wwwdata/reviewed/human_protein_masterlist.csv", 
                        "filename": "human_protein_masterlist.csv"
                    }, 
                    {
                        "access_time": "", 
                        "sha1_chksum": "", 
                        "uri": "https://data.glygen.org/ln2wwwdata/source/human_protein_xref_pharos.csv", 
                        "filename": "human_protein_xref_pharos.csv"
                    }
                ], 
                "output_list": [
                    {
                        "access_time": "", 
                        "uri": "https://data.glygen.org/ln2wwwdata/reviewed/protein_glygen_pharos_xref_mapping.csv", 
                        "filename": "protein_glygen_pharos_xref_mapping.csv"
                    }
                ], 
                "prerequisite": [], 
                "description": "The script loads the human UniProtKB GlyGen accessions that mapped to the Pharos protein accessions along with the GlyGen protein accession URL and extracts it into csv dataset file. "
            }
        ]
    }, 
    "execution_domain": {
        "external_data_endpoints": [], 
        "environment_variables": {}, 
        "script_driver": "manual", 
        "software_prerequisites": [
            {
                "version": "2.7.5", 
                "name": "Python", 
                "uri": {
                    "access_time": "", 
                    "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-species-agnostic-dataset.py", 
                    "filename": "make-species-agnostic-dataset.py"
                }
            }
        ]
    }, 
    "error_domain": {
        "empirical_error": {}, 
        "algorithmic_error": {}
    }, 
    "parametric_domain": [
        {
            "step": "", 
            "value": "", 
            "param": ""
        }
    ], 
    "usability_domain": [
        "The GlyGen-Pharos Cross-references Mapping dataset contains GlyGen's human [taxid:9606] UniProtKB accessions and corresponding GlyGen URLs. The dataset is derived from 2019-09 UniProtKB release.", 
        "The dataset is created for Pharos database to crosslink their protein page links to GlyGen protein pages.", 
        "The log file contains the Pharos protein accessions that could not be mapped to GlyGen protein accessions"
    ], 
    "_id": {
        "$oid": "5e87f166f5c47b1a1a2df9a2"
    }, 
    "bco_id": "http://data.glygen.org/GLY_000464"
}
