{
    "object_id": "https://biocomputeobject.org/GLY_000888/v-2.5.1",
    "spec_version": "1.3.0",
    "provenance_domain": {
        "name": "Human Glycosylation Sites (EMBL)",
        "version": "v-2.5.1",
        "license": "https://creativecommons.org/licenses/by/4.0/",
        "modified": "2024-05-21T13:23:51.724Z",
        "contributors": [
            {
                "name": "Jeet Vora",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "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": "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": "Clement Potel",
                "affiliation": "EMBL",
                "email": "clement.potel@embl.de",
                "contribution": [
                    "contributedBy"
                ]
            },
            {
                "name": "Mira Burtscher",
                "affiliation": "EMBL",
                "email": "mira.burtscher@embl.de",
                "contribution": [
                    "contributedBy"
                ]
            },
            {
                "name": "Martin Garrido",
                "affiliation": "EMBL",
                "email": "martin.garrido@embl.de",
                "contribution": [
                    "contributedBy"
                ]
            },
            {
                "name": "Kate Warner",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "k.warner1@gwu.edu"
            }
        ]
    },
    "usability_domain": [
        "The Human Glycosylation Sites (EMBL) dataset contains glycosylation information (mainly for N-glycosylation) on Homo sapiens (taxid:9606) cell line (HelaK and HEK293T) proteins. The data is submitted by Dr. Clement Potel, Dr. Mira Burtscher and Dr. Martin Garrido, of the Genome Biology Unit, EMBL based on the publication 'Deep quantitative glycoproteomics reveals gut microbiome induced remodeling of the brain glycoproteome.' GlyTouCan accessions were mapped by the GlyGen team to the glycan compositions provided by the authors. If you use this dataset please provide proper attribution to the data source [1], EMBL and GlyGen.",
        "References: 1) Potel C, Burtscher ML, Garrido-Rodriguez M, Brauer-Nikonow A, Becher I, Typas A, Zimmermann M, Savitski M. Deep quantitative glycoproteomics reveals gut microbiome induced remodeling of the brain glycoproteome. bioRxiv 2023.09.13.557529; doi:10.1101/2023.09.13.557529."
    ],
    "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": "Glycosylation Sites",
                        "category_name": "tag"
                    },
                    {
                        "category_value": "csv",
                        "category_name": "file_type"
                    },
                    {
                        "category_value": "Reviewed",
                        "category_name": "status"
                    },
                    {
                        "category_value": "internal",
                        "category_name": "scope"
                    }
                ]
            }
        }
    ],
    "description_domain": {
        "pipeline_steps": [
            {
                "name": "make-proteoform-dataset.py",
                "step_number": 1,
                "input_list": [
                    null
                ],
                "prerequisite": [],
                "output_list": []
            }
        ]
    },
    "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": {}
    },
    "parametric_domain": [
        {
            "value": ""
        }
    ],
    "io_domain": {
        "output_subdomain": [
            {
                "mediatype": "csv",
                "uri": {
                    "filename": "human_proteoform_glycosylation_sites_embl.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.5.1/reviewed/human_proteoform_glycosylation_sites_embl.csv"
                }
            },
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.5.1/reviewed/human_proteoform_glycosylation_sites_embl.stat.csv",
                    "filename": "human_proteoform_glycosylation_sites_embl.stat.csv"
                }
            }
        ],
        "input_subdomain": [
            {
                "uri": {
                    "filename": "n_sequon_info.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.5.1/misc/n_sequon_info.csv"
                }
            },
            {
                "uri": {
                    "filename": "glygen_upload.csv",
                    "uri": "https:/data.glygen.org/ln2downloads/embl/2024_02_09/glygen_upload.csv"
                }
            },
            {
                "uri": {
                    "filename": "names.tsv",
                    "uri": "https:/data.glygen.org/ln2downloads/glytoucan/2024_05_01/export/names.tsv"
                }
            }
        ]
    },
    "etag": "0d454e0e41828af939bd0e22f521709f"
}