{
    "object_id": "https://biocomputeobject.org/GLY_000289/v-1.11.2",
    "spec_version": "1.3.0",
    "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": "InChI Sequence format",
                        "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 Sequences InChI",
        "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": "Karina Martinez",
                "email": "karinamartinez@gwu.edu"
            },
            {
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "name": "Amanda Bell",
                "email": "amandab2140@gwu.edu"
            },
            {
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "name": "Jonathon Keeney",
                "email": "keeneyjg@gwu.edu"
            },
            {
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "name": "Millicent Quartey",
                "email": "mquartey@email.gwu.edu"
            }
        ],
        "created": "2019-11-13T16:26:03.109539",
        "modified": "2020-03-16T13:17:30.653608",
        "version": "v-1.11.2"
    },
    "description_domain": {
        "keywords": [
            "Glycan",
            "GlyTouCan Accession",
            "InChI"
        ],
        "pipeline_steps": [
            {
                "name": "make-species-agnostic-dataset.py",
                "step_number": 1,
                "version": "",
                "input_list": [
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-1.11.2/source/inchi_sdf.tar.gz",
                        "filename": "inchi_sdf.tar.gz"
                    }
                ],
                "output_list": [
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-1.11.2/source/glycan_sequences_inchi.csv",
                        "filename": "glycan_sequences_inchi.csv"
                    }
                ],
                "prerequisite": [],
                "description": "The GlyTouCan Ac. to PubChem CID mapping file was used download .sdf PubChem compound files from PubChem CID ftp and a python script was used to extract the values for headers-<PUBCHEM_COMPOUND_CID> , <PUBCHEM_IUPAC_INCHI>, <PUBCHEM_IUPAC_INCHIKEY>  from the sdf files to generate the final output file."
            }
        ]
    },
    "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"
                }
            }
        ]
    },
    "parametric_domain": [
        {
            "step": "",
            "param": "",
            "value": ""
        }
    ],
    "usability_domain": [
        "The dataset provides the InChI key and glycan sequences in InChI format  for the associated glycans (GlyTouCan Accession) based on the GlyTouCan to PubChem CID mapping (https://data.glygen.org/GLYDS000281). The listed GlyTouCan Accessions are derived from the GlyTouCan accession master list (https://data.glygen.org/GLYDS000144). Source database: https://pubchem.ncbi.nlm.nih.gov/; https://glytoucan.org/"
    ],
    "etag": "698f439dc577c398fb993f92ff26a6c5",
    "_id": {
        "$oid": "5dcc74ebf5c47b2a8119f909"
    },
    "io_domain": {
        "input_subdomain": [
            {
                "uri": {
                    "uri": "https://data.glygen.org//ln2data/releases/data/v-1.11.2/source/inchi_sdf.tar.gz",
                    "filename": "inchi_sdf.tar.gz"
                }
            }
        ],
        "output_subdomain": [
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "http://data.glygen.org/ln2data/releases/data/v-1.11.2/reviewed/glycan_sequences_inchi.csv",
                    "filename": "glycan_sequences_inchi.csv"
                }
            },
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "http://data.glygen.org/ln2data/releases/data/v-1.11.2/reviewed/glycan_sequences_inchi.stat.csv",
                    "filename": "glycan_sequences_inchi.stat.csv"
                }
            }
        ]
    }
}