{
    "io_domain": {
        "input_subdomain": [
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2data/releases/data/v-1.4.5/source/enz.tsv",
                    "filename": "enz.tsv"
                }
            },
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2data/releases/data/v-1.4.5/source/mapped_n_glycans.csv",
                    "filename": "mapped_n_glycans.csv"
                }
            }
        ],
        "output_subdomain": [
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "http://data.glygen.org/ln2data/releases/data/v-1.4.5/reviewed/human_glycan_enzyme.csv",
                    "filename": "human_glycan_enzyme.csv"
                }
            },
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "http://data.glygen.org/ln2data/releases/data/v-1.4.5/reviewed/human_glycan_enzyme.stat.csv",
                    "filename": "human_glycan_enzyme.stat.csv"
                }
            }
        ]
    },
    "spec_version": "1.3.0",
    "object_id": "https://biocomputeobject.org/GLY_000064/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": "Homo sapiens",
                        "category_name": "species"
                    },
                    {
                        "category_value": "Glycan",
                        "category_name": "molecule"
                    },
                    {
                        "category_value": "Biosynthetic Enzyme",
                        "category_name": "tags"
                    },
                    {
                        "category_value": "Reviewed",
                        "category_name": "status"
                    },
                    {
                        "category_value": "csv",
                        "category_name": "file_type"
                    },
                    {
                        "category_value": "internal",
                        "category_name": "scope"
                    }
                ]
            }
        }
    ],
    "provenance_domain": {
        "license": "https://creativecommons.org/licenses/by/4.0/",
        "name": "Human proteins-Biosynthetic Enzyme",
        "contributors": [
            {
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "name": "Rahi Navelkar",
                "email": "rsn13@gwu.edu"
            },
            {
                "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": "Xavier Holmes",
                "email": "xavierh@email.gwu.edu"
            },
            {
                "affiliation": "University of Georgia",
                "contribution": [
                    "authoredBy"
                ],
                "name": "Will York",
                "email": "will@ccrc.uga.edu"
            },
            {
                "affiliation": "University of Georgia",
                "contribution": [
                    "authoredBy"
                ],
                "name": "Allison Nairn",
                "email": "avnairn@uga.edu"
            },
            {
                "affiliation": "Georgetown University",
                "contribution": [
                    "authoredBy"
                ],
                "name": "Nathan Edwards",
                "email": "nje5@georgetown.edu"
            }
        ],
        "created": "2019-01-23T17:21:05.928353",
        "modified": "2019-09-11T10:26:56.854004",
        "version": "v-1.4.5"
    },
    "description_domain": {
        "keywords": [
            "Glycan",
            "GlyTouCan Accession",
            "Enzyme"
        ],
        "pipeline_steps": [
            {
                "name": "make_glycan_dataset.py",
                "step_number": 1,
                "version": "",
                "input_list": [
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-1.4.5/source/mapped_n_glycans.csv",
                        "filename": "mapped_n_glycans.csv"
                    },
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-1.4.5/source/enz.tsv",
                        "filename": "enz.tsv"
                    }
                ],
                "output_list": [
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-1.4.5/source/human_glycan_enzyme.csv",
                        "filename": "human_glycan_enzyme.csv"
                    }
                ],
                "prerequisite": [],
                "description": "The input files were downloaded from the authors and a python script was used to extract columns- GlyTouCanAccession,Enzyme_UniProtAccession, Enzyme_GeneName,Enzyme_GeneID,Enzyme_TaxonomyID from input file 1 and columns- glycan_ID,residue,residue_ID,parent_ID from input file 2 . The script additionally maps the UniProtKB accession to UniProtKB canonical accessions and extracts rows where UniProtKB canonical accession has taxid=9606"
            }
        ]
    },
    "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"
                }
            }
        ]
    },
    "error_domain": {
        "empirical_error": {},
        "algorithmic_error": {},
        "emperical_error": {}
    },
    "parametric_domain": [
        {
            "step": "",
            "value": "",
            "param": ""
        }
    ],
    "usability_domain": [
        "The dataset provides information on biosynthetic enzymes (UniProtKB Ac.) associated to the glycan structures (GlyTouCan Accessions). The enzymes listed are human proteins that are associated with the canonical monosaccharides. The association of a glycan structure to the canonical monosaccharides are based on the GlycO tree model. The glycans listed in the database may or may not have a human species (TaxID:9606) annotation from the source (GlyTouCan or UniCarbKB) database. Please note that a human biosynthetic enzyme association should not be considered as a human species annotation for that glycan."
    ],
    "etag": "",
    "_id": {
        "$oid": "5c48e8d1f5c47b65ae4b5d2b"
    }
}