{
    "object_id": "https://biocomputeobject.org/GLY_000809/v-2.3.1",
    "etag": "13126d72c27300dfcc42c94804b48fc4",
    "spec_version": "1.3.0",
    "provenance_domain": {
        "name": "Yeast Phosphorylation Sites (iPTMNet)",
        "version": "v-2.3.1",
        "created": "2022-09-06T09:56:04.354720",
        "modified": "2023-09-11T20:47:35",
        "contributors": [
            {
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "name": "Jeet Vora",
                "email": "jeetvora@gwu.edu"
            },
            {
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "name": "Karina Martinez",
                "email": "karinamartinez@gwu.edu"
            },
            {
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "name": "Robel Kahsay",
                "email": "rykahsay@gwu.edu"
            },
            {
                "name": "Jingyue Wu",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "jingyue.wu@email.gwu.edu"
            },
            {
                "name": "Jingyue Wu",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "jingyue.wu@email.gwu.edu"
            }
        ],
        "license": "https://creativecommons.org/licenses/by/4.0/"
    },
    "usability_domain": [
        "This dataset contains yeast [taxid:559292] phosphorylation sites from iPTMNet database. The protein and sites are mapped to UniProtKB canonical accessions and fasta sequence. If you use this dataset please provide proper attribution to iPTMnet-GlyGen."
    ],
    "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": "Saccharomyces cerevisiae S288C",
                        "category_name": "species"
                    },
                    {
                        "category_value": "Proteoform",
                        "category_name": "molecule"
                    },
                    {
                        "category_value": "Phosphorylation Sites (iPTMnet)",
                        "category_name": "tag"
                    },
                    {
                        "category_value": "csv",
                        "category_name": "file_type"
                    },
                    {
                        "category_value": "Reviewed",
                        "category_name": "status"
                    },
                    {
                        "category_value": "internal",
                        "category_name": "scope"
                    }
                ]
            }
        }
    ],
    "description_domain": {
        "keywords": [
            "iPTMnet",
            "Phosphorylation sites"
        ],
        "pipeline_steps": [
            {
                "step_number": 0,
                "name": "make-proteoform-dataset.py",
                "description": "The script extracts the phosphorylation information from the source dataset as per taxid and extracts as a csv dataset. The protein accession and sites are mapped to the UniProtKB canonical accession and fasta sequence.",
                "version": "1",
                "prerequisite": [],
                "input_list": [],
                "output_list": []
            }
        ]
    },
    "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": {}
    },
    "parametric_domain": [
        {
            "param": "",
            "value": "",
            "step": ""
        }
    ],
    "io_domain": {
        "input_subdomain": [
            {
                "uri": {
                    "filename": "yeast_protein_masterlist.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.3.1/reviewed/yeast_protein_masterlist.csv"
                }
            },
            {
                "uri": {
                    "filename": "ptm.txt",
                    "uri": "https:/data.glygen.org/ln2downloads/iptmnet/2023_10_13/ptm.txt"
                }
            }
        ],
        "output_subdomain": [
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.3.1/reviewed/yeast_proteoform_phosphorylation_sites_iptmnet.csv",
                    "filename": "yeast_proteoform_phosphorylation_sites_iptmnet.csv"
                }
            },
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.3.1/reviewed/yeast_proteoform_phosphorylation_sites_iptmnet.stat.csv",
                    "filename": "yeast_proteoform_phosphorylation_sites_iptmnet.stat.csv"
                }
            }
        ]
    },
    "error_domain": {
        "empirical_error": {},
        "algorithmic_error": {}
    }
}