{
    "io_domain": {
        "input_subdomain": [
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2data/releases/data/v-1.5.11/source/enz.tsv",
                    "filename": "enz.tsv"
                }
            },
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2data/releases/data/v-1.5.11/source/mapped_n_glycans.csv",
                    "filename": "mapped_n_glycans.csv"
                }
            }
        ],
        "output_subdomain": [
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "http://data.glygen.org/ln2data/releases/data/v-1.5.11/reviewed/glycan_enzyme.csv",
                    "filename": "glycan_enzyme.csv"
                }
            },
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "http://data.glygen.org/ln2data/releases/data/v-1.5.11/reviewed/glycan_enzyme.stat.csv",
                    "filename": "glycan_enzyme.stat.csv"
                }
            }
        ]
    },
    "spec_version": "1.3.0",
    "object_id": "https://biocomputeobject.org/GLY_000284/v-1.5.11",
    "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": "",
                        "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": "Glycan 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": "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": "2019-11-13T13:19:08.883378",
        "modified": "2019-11-14T16:27:18.709811",
        "version": "v-1.5.11"
    },
    "description_domain": {
        "keywords": [
            "Glycan",
            "GlyTouCan Accession",
            "Enzyme"
        ],
        "pipeline_steps": [
            {
                "name": "make-species-agnostic-dataset.py",
                "step_number": 1,
                "version": "",
                "input_list": [
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-1.5.11/source/mapped_n_glycans.csv",
                        "filename": "mapped_n_glycans.csv"
                    },
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-1.5.11/source/enz.tsv",
                        "filename": "enz.tsv"
                    }
                ],
                "output_list": [
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-1.5.11/source/glycan_enzyme.csv",
                        "filename": "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 or Taxid=10090"
            }
        ]
    },
    "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-species-agnostic-dataset.py",
                    "filename": "make-species-agnostic-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 or mouse 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 (TaxID:9606) or mouse (TaxID:10090) species annotation from the source (GlyTouCan or UniCarbKB) database. Please note that a human or mouse biosynthetic enzyme association should not be considered as a human or mouse species annotation for that glycan."
    ],
    "etag": "efa52d8d038b4becd3239706164d757e",
    "_id": {
        "$oid": "5dcc491cf5c47b20a5acac95"
    }
}