{
    "io_domain": {
        "input_subdomain": [
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2data/releases/data/v-1.4.5/source/taxa.tsv",
                    "filename": "taxa.tsv"
                }
            },
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2data/releases/data/v-1.4.5/reviewed/rat_glycan_enzyme.csv",
                    "filename": "rat_glycan_enzyme.csv"
                }
            },
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2data/releases/data/v-1.4.5/reviewed/rat_proteoform_glycosylation_sites_unicarbkb.csv",
                    "filename": "rat_proteoform_glycosylation_sites_unicarbkb.csv"
                }
            },
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2data/releases/data/v-1.4.5/source/allmotif.tsv",
                    "filename": "allmotif.tsv"
                }
            }
        ],
        "output_subdomain": [
            {
                "mediatype": "",
                "uri": {
                    "uri": "http://data.glygen.org/ln2data/releases/data/v-1.4.5/reviewed/rat_glycan_accession_sources.csv",
                    "filename": "rat_glycan_accession_sources.csv"
                }
            },
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "http://data.glygen.org/ln2data/releases/data/v-1.4.5/reviewed/rat_glycan_accession_sources.stat.csv",
                    "filename": "rat_glycan_accession_sources.stat.csv"
                }
            }
        ]
    },
    "spec_version": "1.3.0",
    "object_id": "https://biocomputeobject.org/GLY_000146/v-1.4.5",
    "extension_domain": [
        {
            "extension_schema": "http://www.w3id.org/biocompute/extension_domain/1.1.0/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": "Reviewed",
                        "category_name": "status"
                    },
                    {
                        "category_value": "Glycan",
                        "category_name": "molecule"
                    },
                    {
                        "category_value": "Rattus norvegicus",
                        "category_name": "species"
                    },
                    {
                        "category_value": "Master List",
                        "category_name": "tags"
                    },
                    {
                        "category_value": "csv",
                        "category_name": "file_type"
                    },
                    {
                        "category_value": "internal",
                        "category_name": "scope"
                    }
                ]
            }
        }
    ],
    "provenance_domain": {
        "license": "https://creativecommons.org/licenses/by/4.0/",
        "name": "Rat GlyTouCan Masterlist",
        "contributors": [
            {
                "affiliation": "George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "name": "Rahi Navelkar",
                "email": "rsn13@gwu.edu"
            },
            {
                "affiliation": "George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "name": "Jeet Vora",
                "email": "jeetvora@gwu.edu"
            },
            {
                "affiliation": "George Washington University",
                "contribution": [
                    "authoredBy"
                ],
                "name": "Robel Kahsay",
                "email": "rykahsay@email.gwu.edu"
            }
        ],
        "created": "2019-07-16T12:40:17.523554",
        "modified": "2019-09-12T10:54:44.251594",
        "version": "v-1.4.5"
    },
    "description_domain": {
        "keywords": [
            "Glycan",
            "GlyTouCan Accession",
            "Master List"
        ],
        "pipeline_steps": [
            {
                "name": "python",
                "step_number": 0,
                "version": "",
                "input_list": [
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-1.4.5/source/taxa.tsv",
                        "filename": "taxa.tsv"
                    },
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-1.4.5/reviewed/rat_glycan_enzyme.csv",
                        "filename": "rat_glycan_enzyme.csv"
                    },
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-1.4.5/reviewed/rat_proteoform_glycosylation_sites_unicarbkb.csv",
                        "filename": "rat_proteoform_glycosylation_sites_unicarbkb.csv"
                    },
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-1.4.5/source/allmotif.tsv",
                        "filename": ""
                    }
                ],
                "output_list": [],
                "prerequisite": [],
                "description": "The output file was generated by appending the GlyTouCan accessions from 4 input files. The source column represents the category number based on the source input files: category_I : (Input file 1; where \"Taxid:9606\") The Glycan has a Rat [taxid:10116] (Rattus Norvegicus) species annotation from GlyTouCan and/UniCarbKB database. category_II: (Input file 2) The Glycan is synthesized by a Rat [taxid:10116] (Rattus Norvegicus) protein. category_III: (Input file 3) The Glycan is known to be attached to a Rat [taxid:10116] (Rattus Norvegicus) protein. category_IV: (Input file 4) The glycan is a motif of the glycan from categories I, II, III."
            }
        ]
    },
    "execution_domain": {
        "external_data_endpoints": [],
        "environment_variables": {},
        "script_driver": "manual",
        "software_prerequisites": [
            {
                "version": "2.7.5",
                "name": "Python",
                "uri": {
                    "uri": "https://www.python.org/download/releases/2.7.5/"
                }
            }
        ],
        "script": [
            {
                "uri": {
                    "uri": "https://github.com/GW-HIVE/glygen-backend-integration/blob/master/pipeline/integrator/make-glycan-dataset.py",
                    "filename": "make-glycan-dataset.py"
                }
            }
        ]
    },
    "parametric_domain": [
        {
            "step": "",
            "value": "",
            "param": ""
        }
    ],
    "usability_domain": [
        "The dataset contains the list of GlyTouCan accessions considered to be the GlyGen Rat GlyTouCan Master list. The inclusion criteria are provided by the category number associated with the glycan (GlyTouCan Accession). Category_ I: The Glycan has a Rat [taxid:10116] (Rattus norvegicus) species annotation from GlyTouCan and/or UniCarbKB database. category_II:  The Glycan is associated to a Rat [taxid:10116] (Rattus norvegicus) biosynthetic enzyme. category_III: The Glycan is known to be attached to a Rat [taxid:10116] (Rattus norvegicus) protein. category_IV: The glycan is a motif of the glycan from categories I, II, III."
    ],
    "etag": "",
    "_id": {
        "$oid": "5d2dfdf1f5c47b4fd6db45e2"
    }
}