{
    "object_id": "https://biocomputeobject.org/GLY_000025/v-1.0.13",
    "license": "https://creativecommons.org/licenses/by/4.0/",
    "contributors": [
        {
            "orcid": "",
            "affiliation": "The George Washington University",
            "contribution": [
                "createdBy"
            ],
            "name": "Jeet Vora",
            "email": "jeetvora@gwu.edu"
        },
        {
            "orcid": "",
            "affiliation": "The George Washington University",
            "contribution": [
                "createdBy"
            ],
            "name": "Rahi Navelkar",
            "email": "rsn13@gwu.edu"
        },
        {
            "orcid": "",
            "affiliation": "The George Washington University",
            "contribution": [],
            "name": "Robel Kahsay",
            "email": "rykahsay@gwu.edu"
        },
        {
            "orcid": "",
            "affiliation": "EMBL-EBI-UniProt",
            "contribution": [
                "contributedBy"
            ],
            "name": "Maria Martin",
            "email": "martin@ebi.ac.uk"
        },
        {
            "orcid": "",
            "affiliation": "EMBL-EBI-UniProt",
            "contribution": [
                "contributedBy"
            ],
            "name": "Leyla Garcia",
            "email": "ljgarcia@ebi.ac.uk"
        },
        {
            "orcid": "",
            "affiliation": "EMBL-EBI-UniProt",
            "contribution": [
                "contributedBy"
            ],
            "name": "Preethi Vasudev",
            "email": "pvasudev@ebi.ac.uk"
        }
    ],
    "created": "2019-01-23T17:21:05.892318",
    "spec_version": "1.3.0",
    "modified": "2019-02-21T11:46:03.541450",
    "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"
                    }
                ]
            }
        }
    ],
    "io_domain": {
        "input_subdomain": [
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2data/releases/data/v-1.0.13/source/uniprot-proteome-9606-homo-sapiens-2018-11.nt",
                    "filename": "uniprot-proteome-9606-homo-sapiens-2018-11.nt"
                }
            },
            {
                "uri": {
                    "uri": " https://data.glygen.org/ln2data/releases/data/v-1.0.13/reviewed/human_protein_idmapping.csv",
                    "filename": "human_protein_idmapping.csv"
                }
            }
        ],
        "output_subdomain": [
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https://data.glygen.org/ln2data/releases/data/v-1.0.13/reviewed/human_protein_glycohydrolase.csv",
                    "filename": "human_protein_glycohydrolase.csv"
                }
            },
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https://data.glygen.org/ln2data/releases/data/v-1.0.13/reviewed/human_protein_glycohydrolase.stat.csv",
                    "filename": "human_protein_glycohydrolase.stat.csv"
                }
            }
        ]
    },
    "provenance_domain": {
        "license": "https://creativecommons.org/licenses/by/4.0/",
        "name": "Human Glycohydrolases",
        "contributors": [
            {
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "name": "Jeet Vora",
                "email": "jeetvora@gwu.edu"
            },
            {
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "name": "Rahi Navelkar",
                "email": "rsn13@gwu.edu"
            },
            {
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "name": "Robel Kahsay",
                "email": "rykahsay@gwu.edu"
            }
        ],
        "created": "2019-01-23T17:21:05.892318",
        "modified": "2019-04-24T12:29:07.770672",
        "version": "v-1.0.13"
    },
    "description_domain": {
        "keywords": [
            "glycohydrolase",
            "glycosidase",
            "CAZy"
        ],
        "pipeline_steps": [
            {
                "name": "-",
                "step_number": 1,
                "version": "1.0",
                "input_list": [
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-1.0.13/source/uniprot-proteome-9606-homo-sapiens-2018-11.nt",
                        "filename": "uniprot-proteome-9606-homo-sapiens-2018-11.nt"
                    },
                    {
                        "uri": " https://data.glygen.org/ln2data/releases/data/v-1.0.13/reviewed/human_protein_idmapping.csv",
                        "filename": "human_protein_idmapping.csv"
                    }
                ],
                "output_list": [
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-1.0.13/reviewed/human_protein_glycohydrolase.csv",
                        "filename": "human_protein_glycohydrolase.csv"
                    }
                ],
                "prerequisite": [],
                "description": "Using UniProt advanced search infrastructure three searches were performed UniProtKB/Swiss-Prot (accessed 07/15/2018) using the query \u201ckeyword:Glycosidase [KW-0326]\u201d AND reviewed:yes AND organism: \u201cHomo sapiens (Human) [9606]\u201d, \u201cglycosyl hydrolase\u201d AND reviewed:yes AND organism: \u201cHomo sapiens (Human) [9606]\u201d and \u201cglycoside hydrolase\u201d AND reviewed:yes AND organism: \u201cHomo sapiens (Human) [9606]\u201d, to obtain the list of all reviewed human glycosidase enzymes. For each entry, the UniProtKB Accession, protein name, and gene name (primary) and e.c number were retrieved. The protein entries retrieved from these three queries were reviewed in CAZy, Pfam, InterPro and Brenda. Entries that did not have any association to be glycohydrolase were filtered out. Subunits or inactive proteins were filtered out. The entries from the three lists were merged to get a comprehensive list of glycohydrolases. The UniProtKB accessions were finally mapped to UniProtKB canonicals"
            }
        ]
    },
    "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": "",
                    "filename": ""
                }
            }
        ]
    },
    "version": "1.0.5",
    "error_domain": {
        "empirical_error": {},
        "algorithmic_error": {},
        "emperical_error": {}
    },
    "parametric_domain": [
        {
            "step": "",
            "value": "",
            "param": ""
        }
    ],
    "usability_domain": [
        "The Human Glycohydrolases dataset contains a list of human [taxid:9606] glycohydrolases derived from UniProtKB database using advanced search and other databases including CAZy, InterPro and Pfam. ",
        "If you use this dataset please provide proper attribution to EMBL-EBI-UniProt and GlyGen"
    ],
    "etag": "7dc8563a2f9ed69311a5417700519cbc",
    "_id": {
        "$oid": "5c48e8d1f5c47b65ae4b5d04"
    },
    "name": "Human Glycohydrolases"
}