{
    "provenance_domain": {
        "name": "Glycan PubChem Xref Status",
        "version": "v-2.9.1",
        "created": "2022-09-06T16:07:55.305836",
        "modified": "2024-05-17T11:24:09.560Z",
        "contributors": [
            {
                "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": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "name": "Karina Martinez",
                "email": "karinamartinez@gwu.edu"
            },
            {
                "name": "Luke Johnson",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "luke.johnson@email.gwu.edu"
            },
            {
                "name": "Urnisha Bhuiyan",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "sbhuiyan28@email.gwu.edu"
            },
            {
                "name": "Cyrus Au-Yeung",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "cyruschauyeung@gwu.edu"
            },
            {
                "name": "Kate Warner",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "k.warner1@gwu.edu"
            }
        ],
        "license": "https://creativecommons.org/licenses/by/4.0/"
    },
    "usability_domain": [
        "The dataset provides cross-references status to PubChem compound (CID) and substance (SID) for the associated glycan (GlyTouCan Accession). The listed GlyTouCan Accessions are derived from the GlyTouCan accession master list (https://data.glygen.org/GLYDS000281). The dataset allows to know whether a given GlyGen GlyTouCan has PubChem mapping or not."
    ],
    "extension_domain": [
        {
            "extension_schema": "https://www.w3id.org/biocompute/extension_domain/1.2.0/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": "Glycan",
                        "category_name": "molecule"
                    },
                    {
                        "category_value": "PubChem Xref Status",
                        "category_name": "tag"
                    },
                    {
                        "category_value": "Reviewed",
                        "category_name": "status"
                    },
                    {
                        "category_value": "csv",
                        "category_name": "file_type"
                    },
                    {
                        "category_value": "internal",
                        "category_name": "scope"
                    },
                    {
                        "category_name": "species",
                        "category_value": "Sus scrofa"
                    },
                    {
                        "category_name": "species",
                        "category_value": "Rattus"
                    },
                    {
                        "category_name": "species",
                        "category_value": "Homo sapiens"
                    },
                    {
                        "category_name": "species",
                        "category_value": "Mus musculus"
                    },
                    {
                        "category_name": "species",
                        "category_value": "Saccharomyces cerevisiae"
                    },
                    {
                        "category_name": "species",
                        "category_value": "Drosophila melanogaster"
                    },
                    {
                        "category_name": "species",
                        "category_value": "Rattus norvegicus"
                    },
                    {
                        "category_name": "species",
                        "category_value": "Severe acute respiratory syndrome coronavirus 2"
                    },
                    {
                        "category_name": "species",
                        "category_value": "Sus scrofa domesticus"
                    },
                    {
                        "category_name": "species",
                        "category_value": "Dictyostelium discoideum"
                    },
                    {
                        "category_name": "species",
                        "category_value": "Hepatitis C virus (isolate H77)"
                    },
                    {
                        "category_name": "species",
                        "category_value": "Severe acute respiratory syndrome-related coronavirus"
                    },
                    {
                        "category_name": "species",
                        "category_value": "Hepacivirus C"
                    }
                ]
            }
        }
    ],
    "description_domain": {
        "keywords": [
            "Glycan",
            "GlyTouCan Accession",
            "PubChem",
            "Cross-references",
            "ID/Accession Mapping"
        ],
        "pipeline_steps": [
            {
                "step_number": 1,
                "name": "make-species-agnostic-dataset.py",
                "description": "The script checks all the GlyTouCan accession in masterlist for PubChem mapping. The mapped GlyTouCan accessions to PubChem CID and SID are then extracted in the csv. If the GlyTouCan to PubChem mapping exists then the pubchem_xref_exists is yes",
                "version": "",
                "prerequisite": [],
                "input_list": [],
                "output_list": []
            }
        ]
    },
    "execution_domain": {
        "script": [
            {
                "uri": {
                    "uri": "https://github.com/glygener/glygen-backend-integration/blob/master/dataset-maker/make-species-agnostic-dataset.py",
                    "filename": "make-species-agnostic-dataset.py"
                }
            }
        ],
        "script_driver": "manual",
        "software_prerequisites": [
            {
                "version": "2.7.5",
                "name": "Python",
                "uri": {
                    "uri": "https://www.python.org/download/releases/2.7.5/"
                }
            }
        ],
        "external_data_endpoints": [],
        "environment_variables": {}
    },
    "parametric_domain": [
        {
            "param": "",
            "value": "",
            "step": ""
        }
    ],
    "io_domain": {
        "input_subdomain": [
            {
                "uri": {
                    "filename": "glycan_masterlist.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.9.1/reviewed/glycan_masterlist.csv"
                }
            },
            {
                "uri": {
                    "filename": "glycan_xref_pubchem.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.9.1/reviewed/glycan_xref_pubchem.csv"
                }
            }
        ],
        "output_subdomain": [
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.9.1/reviewed/glycan_pubchem_status.csv",
                    "filename": "glycan_pubchem_status.csv"
                }
            },
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.9.1/reviewed/glycan_pubchem_status.stat.csv",
                    "filename": "glycan_pubchem_status.stat.csv"
                }
            }
        ]
    },
    "object_id": "https://biocomputeobject.org/GLY_000692/v-2.9.1",
    "spec_version": "1.3.0",
    "etag": "f242b19696b9f95f642b53c370778cd0",
    "error_domain": {
        "empirical_error": {},
        "algorithmic_error": {}
    }
}