{
    "object_id": "https://biocomputeobject.org/GLY_000716/v-2.4.1",
    "spec_version": "1.3.0",
    "etag": "aba23f2b90560ba7313d830ab61051d5",
    "provenance_domain": {
        "name": "Human O-Glucosylation Glycosylation Sites",
        "version": "v-2.4.1",
        "created": "2023-02-27T21:41:28.559Z",
        "modified": "2023-03-03T02:07:18.965Z",
        "embargo": {
            "start_time": "",
            "end_time": ""
        },
        "contributors": [
            {
                "name": "Jeet Vora",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy",
                    "curatedBy"
                ],
                "email": "jeetvora@gwu.edu"
            },
            {
                "name": "Karina Martinez",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "karinamartinez@gwu.edu"
            },
            {
                "name": "Robel Kahsay",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "rykahsay@gwu.edu"
            },
            {
                "name": "Daniel Williamson",
                "contribution": [
                    "authoredBy",
                    "contributedBy"
                ],
                "affiliation": "CCRC, UGA",
                "email": "daniel.williamson25@uga.edu",
                "orcid": ""
            },
            {
                "name": "Robert S Haltiwanger",
                "contribution": [
                    "createdAt",
                    "contributedBy"
                ],
                "affiliation": "CCRC, UGA",
                "email": "rhalti@uga.edu",
                "orcid": ""
            },
            {
                "name": "Luke Johnson",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "luke.johnson@email.gwu.edu"
            },
            {
                "name": "Urnisha Bhuiyan",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "sbhuiyan28@email.gwu.edu"
            },
            {
                "name": "Cyrus Au-Yeung",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "cyruschauyeung@gwu.edu"
            },
            {
                "name": "Jingyue Wu",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "jingyue.wu@email.gwu.edu"
            }
        ],
        "license": "https://creativecommons.org/licenses/by/4.0/"
    },
    "usability_domain": [
        "The Human O-Gluc Glycosylation Sites contains human (taxid:9606) O-Glucosylation sites from OGLUT2 and POGLUT3 O - glucosylate multiple EGF repeats in fibrillin - 1, -2, and LTBP1 and promote secretion of fibrillin - 1 [PMID:34411563] (experimental) and based on the consensus sequence the other sites (predicted) were detected using Prosite Prorule. The GlyTouCan ID for O-Gluc is G71142DF and the consensus sequence for O-gluc glycosylation is C3-x-N-T-x-G-S-(FY)-x-C4. The enzymes for o-glucosylation is POGLUT2 and POGLUT3. The dataset also contains Glycosylation stoichiometry: The relative amount (percentage) of a glycoform based on peptide signal intensity determined by mass spectrometry.",
        "The dataset is generated in the lab of Robert S Haltiwanger by Daniel Williamson."
    ],
    "description_domain": {
        "keywords": [
            "",
            "O-Gluc",
            "O-Glucosylation"
        ],
        "platform": [],
        "pipeline_steps": [
            {
                "name": "make-proteoform-dataset.py",
                "version": "",
                "step_number": 1,
                "input_list": [],
                "output_list": [],
                "prerequisite": [],
                "description": "The script maps the UniProtKB accesions to the UniProtKB Canonical accessions and also checks the position and amino acid of O-Gluc modified residue against UniProtKB Canonical sequences."
            }
        ]
    },
    "execution_domain": {
        "script": [
            {
                "uri": {
                    "uri": "https://github.com/glygener/glygen-backend-integration/blob/master/dataset-maker/make-proteoform-dataset.py",
                    "filename": "make-proteoform-dataset.py"
                }
            }
        ],
        "script_driver": "manual",
        "software_prerequisites": [
            {
                "version": "2.7.5",
                "name": "Python",
                "uri": {
                    "uri": "https://www.python.org/download/releases/2.7.5/"
                }
            }
        ],
        "external_data_endpoints": [],
        "environment_variables": {}
    },
    "io_domain": {
        "input_subdomain": [
            {
                "uri": {
                    "filename": "human_proteoform_glycosylation_sites_o-gluc.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.4.1/compiled/human_proteoform_glycosylation_sites_o-gluc.csv"
                }
            },
            {
                "uri": {
                    "filename": "sample_mapping.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.4.1/misc/sample_mapping.csv"
                }
            },
            {
                "uri": {
                    "filename": "cellosaurus.txt",
                    "uri": "https:/data.glygen.org/ln2downloads/cellosaurus/2024_01_31/cellosaurus.txt"
                }
            }
        ],
        "output_subdomain": [
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.4.1/reviewed/human_proteoform_glycosylation_sites_o_gluc.csv",
                    "filename": "human_proteoform_glycosylation_sites_o_gluc.csv"
                }
            },
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.4.1/reviewed/human_proteoform_glycosylation_sites_o_gluc.stat.csv",
                    "filename": "human_proteoform_glycosylation_sites_o_gluc.stat.csv"
                }
            }
        ]
    },
    "parametric_domain": [
        {
            "step": "",
            "param": "",
            "value": ""
        }
    ],
    "extension_domain": [
        {
            "extension_schema": "https://www.w3id.org/biocompute/extension_domain/1.2.0/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": "Proteoform",
                        "category_name": "molecule"
                    },
                    {
                        "category_value": "O-Glucosylation Sites",
                        "category_name": "tag"
                    },
                    {
                        "category_value": "csv",
                        "category_name": "file_type"
                    },
                    {
                        "category_value": "Reviewed",
                        "category_name": "status"
                    },
                    {
                        "category_value": "internal",
                        "category_name": "scope"
                    }
                ]
            }
        }
    ]
}