{
    "object_id": "https://biocomputeobject.org/GLY_000926/v-2.9.1",
    "etag": "a3b2cb61059e67f0f6935cb5250160c5",
    "spec_version": "1.3.0",
    "provenance_domain": {
        "name": "Pig Protein PTM Annotation (UniProtKB)",
        "version": "v-2.9.1",
        "license": "https://creativecommons.org/licenses/by/4.0/",
        "derived_from": [],
        "created": "",
        "modified": "2024-05-17T11:24:18.609Z",
        "review": [],
        "contributors": [
            {
                "name": "Jeet Vora",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "jeetvora@gwu.edu"
            },
            {
                "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": "Jingyue Wu",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "jingyue.wu@email.gwu.edu"
            },
            {
                "name": "Kate Warner",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "k.warner1@gwu.edu"
            }
        ]
    },
    "usability_domain": [
        "The Pig Protein PTM Annotation (UniProtKB) dataset contains Sus scrofa [taxid:9823] PTM annotations for UniProtKB canonical accessions. The annotation is the curated statements of mainly about glycosylation and phosphorylation. If you use this dataset please give proper attribution to UniProtKB and GlyGen."
    ],
    "extension_domain": [
        {
            "dataset_extension": {
                "dataset_categories": [
                    {
                        "category_value": "Sus scrofa",
                        "category_name": "species"
                    },
                    {
                        "category_value": "Protein",
                        "category_name": "molecule"
                    },
                    {
                        "category_value": "PTM annotation",
                        "category_name": "tag"
                    },
                    {
                        "category_value": "csv",
                        "category_name": "file_type"
                    },
                    {
                        "category_value": "Reviewed",
                        "category_name": "status"
                    },
                    {
                        "category_value": "internal",
                        "category_name": "scope"
                    }
                ],
                "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"
        }
    ],
    "description_domain": {
        "keywords": [
            "ptm annotation",
            "glycosylation",
            "phosphorylation"
        ],
        "pipeline_steps": [
            {
                "name": "make-protein-dataset.py",
                "version": "",
                "step_number": 1,
                "input_list": [],
                "output_list": [],
                "prerequisite": [],
                "description": "The script extracts the ptm annotation strings from the triplestore for the UniProtKB canonical sequnces asnd extracts as a dataset in a csv format"
            }
        ]
    },
    "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-protein-dataset.py",
                    "filename": "make-protein-dataset.py"
                }
            }
        ]
    },
    "parametric_domain": [
        {
            "step": "",
            "param": "",
            "value": ""
        }
    ],
    "io_domain": {
        "output_subdomain": [
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.9.1/reviewed/pig_protein_ptm_annotation_uniprotkb.csv",
                    "filename": "pig_protein_ptm_annotation_uniprotkb.csv"
                }
            },
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.9.1/reviewed/pig_protein_ptm_annotation_uniprotkb.stat.csv",
                    "filename": "pig_protein_ptm_annotation_uniprotkb.stat.csv"
                }
            }
        ],
        "input_subdomain": [
            {
                "uri": {
                    "filename": "uniprot-proteome-sus-scrofa.nt",
                    "uri": "https:/data.glygen.org/ln2downloads/ebi/2025_05_20/uniprot-proteome-sus-scrofa.nt"
                }
            }
        ]
    }
}