{
    "io_domain": {
        "output_subdomain": [
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.12.1/reviewed/mouse_proteoform_glycosylation_sites_literature_mining_manually_verified.csv",
                    "filename": "mouse_proteoform_glycosylation_sites_literature_mining_manually_verified.csv"
                }
            },
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.12.1/reviewed/mouse_proteoform_glycosylation_sites_literature_mining_manually_verified.stat.csv",
                    "filename": "mouse_proteoform_glycosylation_sites_literature_mining_manually_verified.stat.csv"
                }
            }
        ],
        "input_subdomain": [
            {
                "uri": {
                    "filename": "proteoform_glycosylation_sites_literature_mining_manually_verified.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.12.1/compiled/proteoform_glycosylation_sites_literature_mining_manually_verified.csv"
                }
            },
            {
                "uri": {
                    "filename": "n_sequon_info.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.12.1/misc/n_sequon_info.csv"
                }
            },
            {
                "uri": {
                    "filename": "sample_mapping.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.12.1/misc/sample_mapping.csv"
                }
            },
            {
                "uri": {
                    "filename": "cellosaurus.txt",
                    "uri": "https:/data.glygen.org/ln2downloads/cellosaurus/2026_06_02/cellosaurus.txt"
                }
            }
        ]
    },
    "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": "Reviewed",
                        "category_name": "status"
                    },
                    {
                        "category_value": "Proteoform",
                        "category_name": "molecule"
                    },
                    {
                        "category_value": "Mus musculus",
                        "category_name": "species"
                    },
                    {
                        "category_value": "glycosylation sites",
                        "category_name": "tag"
                    },
                    {
                        "category_value": "csv",
                        "category_name": "file_type"
                    },
                    {
                        "category_value": "internal",
                        "category_name": "scope"
                    },
                    {
                        "category_name": "file_category",
                        "category_value": "glycosylation_sites_"
                    }
                ]
            }
        }
    ],
    "provenance_domain": {
        "contributors": [
            {
                "affiliation": "George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "name": "Jeet Vora",
                "email": "jeetvora@gwu.edu"
            },
            {
                "affiliation": "George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "name": "Robel Kahsay",
                "email": "rykahsay@email.gwu.edu"
            },
            {
                "affiliation": "University of Delaware",
                "contribution": [
                    "contributedBy"
                ],
                "name": "Vijay Shanker",
                "email": "vijay@udel.edu"
            },
            {
                "affiliation": "Georgetown University",
                "contribution": [
                    "contributedBy",
                    "curatedBy"
                ],
                "name": "Karen Ross",
                "email": "ker25@georgetown.edu"
            },
            {
                "affiliation": "GlyConnect",
                "contribution": [
                    "curatedBy"
                ],
                "name": "Catherine Hayes",
                "email": "hayescat@gmail.com"
            },
            {
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "name": "Karina Martinez",
                "email": "karinamartinez@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": "Kate Warner",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "k.warner1@gwu.edu"
            }
        ],
        "license": "https://creativecommons.org/licenses/by/4.0/",
        "created": "2021-08-02T15:48:30.176124",
        "modified": "2024-05-21T15:35:20.961Z",
        "version": "v-2.12.1",
        "name": "Mouse glycosylation sites [Automatic literature Mining] [Manually verified]"
    },
    "description_domain": {
        "keywords": [
            "proteins",
            "canonical",
            "glycosylation sites",
            "GlyTouCan",
            "Automatic Literature Mining",
            "Manual QC"
        ],
        "pipeline_steps": [
            {
                "name": "make-proteoform-dataset.py",
                "step_number": 1,
                "input_list": [],
                "description": "The input file was downloaded from the provider(s) and QC steps were done to validate the protein and site annotation. Additional QC was done to check wether the annotion (protein +site) is already present in GlyGen or not. Only those rows where extracted where the rows failed the annotation QC i.e where the annotation didn't already exist in GlyGen. Such rows where manually verified by the curators and again only those rows that passed where added to the file. ",
                "prerequisite": [],
                "output_list": []
            }
        ]
    },
    "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/glygener/glygen-backend-integration/blob/master/dataset-maker/make-proteoform-dataset.py",
                    "filename": "make-proteoform-dataset.py"
                }
            }
        ]
    },
    "error_domain": {},
    "parametric_domain": [
        {
            "value": ""
        }
    ],
    "usability_domain": [
        "The Mouse glycosylation sites [Automatic literature Mining] [Manually verified] dataset contains a list of Mus musculus [taxid:10090] proteins with information on glycosylation sites detected from PubMed database (abstracts-only) through Automatic mining tool developed by Dr. Vijay Shanker (University of Delaware). This file contains rows that are detected from automatic literature mining and which failed the automatic QC where the annotation (protein+site) didn't already exist in GlyGen. such rows are then manually verified by curators. The listed protein (UniProtKB) accessions are part of the GlyGen UniProtKB canonical list (https://data.glygen.org/GLY_000007)."
    ],
    "object_id": "https://biocomputeobject.org/GLY_000583/v-2.12.1",
    "spec_version": "1.3.0",
    "etag": "710e7d7b91498baa0d815658f2c3ebc4"
}