{
    "object_id": "https://biocomputeobject.org/GLY_001166/v-2.11.1",
    "spec_version": "1.3.0",
    "etag": "5d44cf407d33eb34a5cb7deb7618d5d4",
    "provenance_domain": {
        "name": "Mouse Proteoform Glycosylation Citations (CarbBank)",
        "version": "v-2.11.1",
        "license": "https://creativecommons.org/licenses/by/4.0/",
        "created": "2024-12-10T21:03:46",
        "modified": "2024-12-10T22:19:00.713Z",
        "contributors": [
            {
                "name": "Jeet Vora",
                "affiliation": "The George Washington University",
                "email": "jeetvora@gwu.edu",
                "contribution": [
                    "createdBy"
                ]
            },
            {
                "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": "Kate Warner",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "k.warner1@gwu.edu"
            }
        ]
    },
    "usability_domain": [
        "The Mouse Proteoform Glycosylation Citations (CarbBank) dataset contains Mus musculus [taxid:10090] publication information of glycoproteins from the Mouse CarbBank Glycosylation Sites dataset (https://data.glygen.org/GLY_001114). The script accesses the NCBI ENTREZ infrastructure to retrieve the publication information of the given PMID from the CarbBank dataset. CarbBank was a pioneering database that cataloged carbohydrate structures and related biochemical information. Although no longer actively maintained, CarbBank laid the groundwork for modern glycoinformatics tools and databases, influencing subsequent efforts to document carbohydrate-related data systematically. Please cite GlyGen and CarbBank if you use this dataset. "
    ],
    "description_domain": {
        "pipeline_steps": [
            {
                "name": "make-proteoform-dataset.py",
                "step_number": 1,
                "input_list": [],
                "description": "The script extracts the PMIDs from the dataset and uses the Entrez API to retrieve the publication details and extracts in a csv dataset",
                "prerequisite": [],
                "output_list": []
            }
        ],
        "keywords": [
            "glycosylation",
            "reported glycosylation",
            "Carbbank"
        ]
    },
    "parametric_domain": [
        {
            "value": ""
        }
    ],
    "io_domain": {
        "input_subdomain": [
            {
                "uri": {
                    "filename": "all_blacklisted_pmids.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.11.1/compiled/all_blacklisted_pmids.csv"
                }
            },
            {
                "uri": {
                    "filename": "mouse_proteoform_glycosylation_sites_carbbank.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.11.1/reviewed/mouse_proteoform_glycosylation_sites_carbbank.csv"
                }
            }
        ],
        "output_subdomain": [
            {
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.11.1/reviewed/mouse_proteoform_citations_glycosylation_sites_carbbank.csv",
                    "filename": "mouse_proteoform_citations_glycosylation_sites_carbbank.csv"
                },
                "mediatype": "csv"
            },
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.11.1/reviewed/mouse_proteoform_citations_glycosylation_sites_carbbank.stat.csv",
                    "filename": "mouse_proteoform_citations_glycosylation_sites_carbbank.stat.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": "Mus musculus",
                        "category_name": "species"
                    },
                    {
                        "category_value": "Proteoform",
                        "category_name": "molecule"
                    },
                    {
                        "category_value": "Protein Citations O-GlcNAc (Carbbank)",
                        "category_name": "tag"
                    },
                    {
                        "category_value": "csv",
                        "category_name": "file_type"
                    },
                    {
                        "category_value": "Reviewed",
                        "category_name": "status"
                    },
                    {
                        "category_value": "internal",
                        "category_name": "scope"
                    }
                ]
            }
        }
    ],
    "error_domain": {}
}