{
    "object_id": "https://biocomputeobject.org/GLY_001495/v-2.12.1",
    "spec_version": "1.3.0",
    "etag": "a58f7a06afc0fc2cb9d6c34f404ebf2b",
    "provenance_domain": {
        "name": "Fruit Fly Glycosylation Sites and Glycans (RCSB PDB)",
        "version": "v-2.12.1",
        "license": "https://creativecommons.org/licenses/by/4.0/",
        "created": "2023-03-14T20:00:08.127Z",
        "modified": "2026-08-18T14:28:54.470Z",
        "contributors": [
            {
                "name": "Jeet Vora",
                "affiliation": "The George Washington University",
                "email": "jeetvora@gwu.edu",
                "contribution": [
                    "createdBy"
                ]
            },
            {
                "name": "Robel Kahsay",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "rykahsay@gwu.edu"
            },
            {
                "name": "Urnisha Bhuiyan",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "sbhuiyan28@email.gwu.edu"
            },
            {
                "name": "Kate Warner",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "k.warner1@gwu.edu"
            }
        ]
    },
    "usability_domain": [
        "The Fruit Fly Glycosylation Sites and Glycans (RCSB PDB) dataset contains Drosophila melanogaster [taxid:7227] glycosylation sites and associated glycans provided by the RCSB PDB database. Data is retrieved using the RCSB PDB Search API (https://search.rcsb.org/#search-api), Data API (https://data.rcsb.org/index.html#data-api) and Sequence Coordinates Server API (https://sequence-coordinates.rcsb.org/#sequence-coordinates-api). If you use this dataset please provide proper attribution to RCSB PDB and GlyGen."
    ],
    "description_domain": {
        "pipeline_steps": [
            {
                "name": "make-proteoform-dataset.py",
                "step_number": 1,
                "input_list": [],
                "description": "The script retrieves data from RCSB PDB in three steps: (1) their RCSB PDB Search API (https://search.rcsb.org/#search-api) is used to retrieve any PDB structures with glycosylation data for this species, followed by (2) their RCSB PDB Data API (https://data.rcsb.org/index.html#data-api) to retrieve data on the glycosylated structures identified in step one, including data on the PDB glycosite position, GlyTouCan IDs and publications, and then (3) the RCSB PDB Sequence Coordinates Server API (https://sequence-coordinates.rcsb.org/#sequence-coordinates-api) to map the chain glycosites to the full UniProtKB protein sequence. The script then extracts the data from the API responses to create the dataset.",
                "prerequisite": [],
                "output_list": []
            }
        ],
        "keywords": [
            "glycosylation",
            "reported glycosylation",
            "UniProtKB",
            "PubMed",
            "DOI",
            "GlyTouCan",
            "RCSB PDB",
            "PDB"
        ]
    },
    "io_domain": {
        "output_subdomain": [
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.12.1/reviewed/fruitfly_proteoform_glycosylation_sites_rcsb_pdb.csv",
                    "filename": "fruitfly_proteoform_glycosylation_sites_rcsb_pdb.csv"
                }
            },
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.12.1/reviewed/fruitfly_proteoform_glycosylation_sites_rcsb_pdb.stat.csv",
                    "filename": "fruitfly_proteoform_glycosylation_sites_rcsb_pdb.stat.csv"
                }
            }
        ],
        "input_subdomain": [
            {
                "uri": {
                    "filename": "n_sequon_info.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.12.1/misc/n_sequon_info.csv"
                }
            },
            {
                "uri": {
                    "filename": "fruitfly_glycosites_rcsb_pdb.csv",
                    "uri": "https:/data.glygen.org/ln2downloads/rcsb_pdb/06_11_2026/fruitfly_glycosites_rcsb_pdb.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": [
        {
            "dataset_extension": {
                "dataset_categories": [
                    {
                        "category_value": "Drosophila melanogaster",
                        "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"
                    },
                    {
                        "category_name": "file_category",
                        "category_value": "glycosylation_sites_"
                    }
                ],
                "additional_license": {
                    "data_license": "https://creativecommons.org/licenses/by/4.0/",
                    "script_license": "https://www.gnu.org/licenses/gpl-3.0.en.html"
                }
            },
            "extension_schema": "https://www.w3id.org/biocompute/extension_domain/1.2.0/dataset/dataset_extension.json"
        }
    ],
    "error_domain": {}
}