{
    "object_id": "https://biocomputeobject.org/GLY_000943/v-2.12.1",
    "spec_version": "1.3.0",
    "etag": "9b4a747f5c682e30fb23c3de3cc5d80f",
    "provenance_domain": {
        "name": "Pig Protein Site Annotation",
        "version": "v-2.12.1",
        "license": "https://creativecommons.org/licenses/by/4.0/",
        "created": "2019-12-01T15:29:48.229274",
        "modified": "2025-10-09T15:07:11.153Z",
        "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": "Maria Martin",
                "affiliation": "EMBL-EBI-UniProt",
                "contribution": [
                    "contributedBy"
                ],
                "email": "martin@ebi.ac.uk"
            },
            {
                "name": "Preethi Vasudev",
                "affiliation": "EMBL-EBI-UniProt",
                "contribution": [
                    "contributedBy"
                ],
                "email": "pvasudev@ebi.ac.uk"
            },
            {
                "name": "Karina Martinez",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "karinamartinez@gwu.edu"
            },
            {
                "name": "Jingyue Wu",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "jingyue.wu@email.gwu.edu"
            },
            {
                "name": "Urnisha Bhuiyan",
                "affiliation": "The George Washington University",
                "email": "sbhuiyan28@email.gwu.edu",
                "contribution": [
                    "createdBy"
                ]
            },
            {
                "name": "Kate Warner",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "k.warner1@gwu.edu"
            }
        ]
    },
    "usability_domain": [
        "The Pig Protein Site Annotation (UniProtKB) dataset contains Sus scrofa [taxid:1758] site annotations for UniProtKB canonical accessions. The dataset contains information about UniProtKB annotations with their amino acid range. If you use this dataset please give proper attribution to UniProtKB and GlyGen."
    ],
    "description_domain": {
        "pipeline_steps": [
            {
                "name": "make-protein-dataset.py",
                "step_number": 1,
                "input_list": [],
                "description": "The script extracts the site annotation triples from the triplestore for the UniProtKB canonical sequnces asnd extracts as a dataset in a csv format",
                "prerequisite": [],
                "output_list": []
            }
        ],
        "keywords": [
            "site annotation",
            "amino acid range",
            "UniProtKB"
        ]
    },
    "io_domain": {
        "output_subdomain": [
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.12.1/reviewed/pig_protein_site_annotation_uniprotkb.csv",
                    "filename": "pig_protein_site_annotation_uniprotkb.csv"
                }
            },
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.12.1/reviewed/pig_protein_site_annotation_uniprotkb.stat.csv",
                    "filename": "pig_protein_site_annotation_uniprotkb.stat.csv"
                }
            }
        ],
        "input_subdomain": [
            {
                "uri": {
                    "filename": "all_blacklisted_pmids.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.12.1/compiled/all_blacklisted_pmids.csv"
                }
            },
            {
                "uri": {
                    "filename": "pig_protein_masterlist.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.12.1/reviewed/pig_protein_masterlist.csv"
                }
            },
            {
                "uri": {
                    "filename": "uniprot-proteome-sus-scrofa.nt",
                    "uri": "https:/data.glygen.org/ln2downloads/ebi/2026_06_13/uniprot-proteome-sus-scrofa.nt"
                }
            }
        ]
    },
    "execution_domain": {
        "script": [
            {
                "uri": {
                    "uri": "https://github.com/glygener/glygen-backend-integration/blob/master/dataset-maker/make-protein-dataset.py",
                    "filename": "make-protein-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": "Sus scrofa",
                        "category_name": "species"
                    },
                    {
                        "category_value": "Protein",
                        "category_name": "molecule"
                    },
                    {
                        "category_value": "site annotation",
                        "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": "annotation"
                    }
                ]
            }
        }
    ],
    "error_domain": {}
}
