{
    "object_id": "https://biocomputeobject.org/GLY_001399/v-2.11.1",
    "spec_version": "1.3.0",
    "etag": "450123ae25a15cd4fb1fa57338d65807",
    "provenance_domain": {
        "name": "SARS-CoV2 Glycosylation Sites [Automatic Literature Mining] [Automatically verified]",
        "version": "v-2.11.1",
        "license": "https://creativecommons.org/licenses/by/4.0/",
        "created": "2025-04-15T14:10:22",
        "modified": "2025-04-15T14:12:00.629Z",
        "contributors": [
            {
                "name": "Jeet Vora",
                "affiliation": "George Washington University",
                "email": "jeetvora@gwu.edu",
                "contribution": [
                    "createdBy"
                ]
            },
            {
                "name": "Robel Kahsay",
                "affiliation": "George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "rykahsay@email.gwu.edu"
            },
            {
                "name": "Vijay Shanker",
                "affiliation": "University of Delaware",
                "contribution": [
                    "contributedBy"
                ],
                "email": "vijay@udel.edu"
            },
            {
                "name": "Karen Ross",
                "affiliation": "Georgetown University",
                "contribution": [
                    "contributedBy"
                ],
                "email": "ker25@georgetown.edu"
            },
            {
                "name": "Karina Martinez",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "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"
            }
        ]
    },
    "usability_domain": [
        "The SARS coronavirus (SARS-CoV-2 or 2019-nCov) Glycosylation Sites [Automatic Literature Mining] [Automatically verified] dataset contains a list of Severe acute respiratory syndrome coronavirus 2 [taxid:2697049] proteins with information on glycosylation sites detected from PubMed database (abstracts-only) through the Automatic mining tool 'GlycoSiteMiner' developed by the GlyGen team (https://github.com/glygener/glycositeminer). The listed protein (UniProtKB) accessions are part of the GlyGen UniProtKB canonical list (https://data.glygen.org/GLY_000434)"
    ],
    "description_domain": {
        "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 whether the annotation (protein +site) is already present in GlyGen or not. Only those rows which passed all the QC filters where added to the output file.",
                "prerequisite": [],
                "output_list": []
            }
        ],
        "keywords": [
            "proteins",
            "canonical",
            "glycosylation sites",
            "GlyTouCan",
            "Automatic Literature Mining"
        ]
    },
    "parametric_domain": [
        {
            "value": ""
        }
    ],
    "io_domain": {
        "output_subdomain": [
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.11.1/reviewed/sarscov2_proteoform_glycosylation_sites_literature_mining.csv",
                    "filename": "sarscov2_proteoform_glycosylation_sites_literature_mining.csv"
                }
            },
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.11.1/reviewed/sarscov2_proteoform_glycosylation_sites_literature_mining.stat.csv",
                    "filename": "sarscov2_proteoform_glycosylation_sites_literature_mining.stat.csv"
                }
            }
        ],
        "input_subdomain": [
            {
                "uri": {
                    "filename": "n_sequon_info.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.11.1/misc/n_sequon_info.csv"
                }
            },
            {
                "uri": {
                    "filename": "sample_mapping.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.11.1/misc/sample_mapping.csv"
                }
            },
            {
                "uri": {
                    "filename": "cellosaurus.txt",
                    "uri": "https:/data.glygen.org/ln2downloads/cellosaurus/2026_01_13/cellosaurus.txt"
                }
            },
            {
                "uri": {
                    "filename": "glycosylation_sites_GlycoSiteMiner_1.0.csv",
                    "uri": "https:/data.glygen.org/ln2downloads/lit_min/2021_04_08/glycosylation_sites_GlycoSiteMiner_1.0.csv"
                }
            },
            {
                "uri": {
                    "filename": "glycosylation_sites_RLIMS-G_1.0.csv",
                    "uri": "https:/data.glygen.org/ln2downloads/lit_min/2021_04_08/glycosylation_sites_RLIMS-G_1.0.csv"
                }
            },
            {
                "uri": {
                    "filename": "glycosylation_sites_GlycoSiteMiner_2.0.csv",
                    "uri": "https:/data.glygen.org/ln2downloads/lit_min/2024_09_30/glycosylation_sites_GlycoSiteMiner_2.0.csv"
                }
            },
            {
                "uri": {
                    "filename": "glycosylation_sites_GlycoSiteMiner_4.0.csv",
                    "uri": "https:/data.glygen.org/ln2downloads/lit_min/2025_11_05/glycosylation_sites_GlycoSiteMiner_4.0.csv"
                }
            }
        ]
    },
    "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": {}
    },
    "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": "Severe acute respiratory syndrome coronavirus 2",
                        "category_name": "species"
                    },
                    {
                        "category_value": "glycosylation sites",
                        "category_name": "tag"
                    },
                    {
                        "category_value": "csv",
                        "category_name": "file_type"
                    },
                    {
                        "category_value": "internal",
                        "category_name": "scope"
                    }
                ]
            }
        }
    ],
    "error_domain": {}
}