{
    "object_id": "https://biocomputeobject.org/GLY_000709/v-2.0.3",
    "spec_version": "1.3.0",
    "etag": "86ddba0f801a931bb517c9ba7aba72a9",
    "provenance_domain": {
        "name": "Mouse O-GlcNAc Glycosylation Sites (O-GlcNAc Atlas)",
        "version": "v-2.0.3",
        "created": "2023-02-24T16:22:06.506Z",
        "modified": "2023-03-06T20:14:30.429Z",
        "embargo": {
            "start_time": "",
            "end_time": ""
        },
        "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": "Jungfeng Ma",
                "contribution": [
                    "contributedBy"
                ],
                "affiliation": "Georgetown University Medical Center, Georgetown University",
                "email": "junfeng.ma@georgetown.edu ",
                "orcid": ""
            }
        ],
        "license": "https://creativecommons.org/licenses/by/4.0/"
    },
    "usability_domain": [
        "The Mouse O-GlcNAc Glycosylation Sites (O-GlcNAc Atlas) contains mouse (taxid:10090) O-GlcNAc modifications on Serine and Threonine extracted from the publication and mapped to the UniProtKB canonical accession and fasta sequence. The GlyTouCan ID for the O-GlcNAc is G49108TO."
    ],
    "description_domain": {
        "keywords": [
            "",
            "O-GlcNAc",
            "Serine"
        ],
        "platform": [],
        "pipeline_steps": [
            {
                "name": "make-proteoform-dataset.py",
                "version": "",
                "step_number": 1,
                "input_list": [
                    {
                        "uri": "",
                        "filename": ""
                    }
                ],
                "output_list": [
                    {
                        "uri": "",
                        "filename": ""
                    }
                ],
                "prerequisite": [],
                "description": "The script maps the UniProtKB accesions to the UniProtKB Canonical accessions and also checks the position and amino acid of O-GlcNAc modified residue against UniProtKB Canonical sequences. The script also maps the sample type to correct sample term id and sample term name. "
            }
        ]
    },
    "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": {}
    },
    "io_domain": {
        "input_subdomain": [
            {
                "uri": {
                    "filename": "mouse_protein_info_uniprotkb.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.0.3/reviewed/mouse_protein_info_uniprotkb.csv"
                }
            },
            {
                "uri": {
                    "filename": "mouse_protein_masterlist.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.0.3/reviewed/mouse_protein_masterlist.csv"
                }
            },
            {
                "uri": {
                    "filename": "mouse_protein_recnames.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.0.3/reviewed/mouse_protein_recnames.csv"
                }
            },
            {
                "uri": {
                    "filename": "cellline_mapping.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.0.3/misc/cellline_mapping.csv"
                }
            },
            {
                "uri": {
                    "filename": "sample_mapping.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.0.3/misc/sample_mapping.csv"
                }
            },
            {
                "uri": {
                    "filename": "tissue_mapping.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.0.3/misc/tissue_mapping.csv"
                }
            },
            {
                "uri": {
                    "filename": "ambiguous_sites_version_2.0.csv",
                    "uri": "/data/shared/glyds/downloads/atlas_oglcnac/2022_10_20/ambiguous_sites_version_2.0.csv"
                }
            },
            {
                "uri": {
                    "filename": "unambiguous_sites_version_2.0.csv",
                    "uri": "/data/shared/glyds/downloads/atlas_oglcnac/2022_10_20/unambiguous_sites_version_2.0.csv"
                }
            },
            {
                "uri": {
                    "filename": "cellosaurus.txt",
                    "uri": "/data/shared/glyds/downloads/cellosaurus/2020-09-15/cellosaurus.txt"
                }
            }
        ],
        "output_subdomain": [
            {
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.0.3/reviewed/mouse_proteoform_glycosylation_sites_oglcnac_atlas.csv",
                    "filename": "mouse_proteoform_glycosylation_sites_oglcnac_atlas.csv",
                    "sha1_checksum": ""
                },
                "mediatype": "csv"
            },
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.0.3/reviewed/mouse_proteoform_glycosylation_sites_oglcnac_atlas.stat.csv",
                    "filename": "mouse_proteoform_glycosylation_sites_oglcnac_atlas.stat.csv"
                }
            }
        ]
    },
    "parametric_domain": [
        {
            "step": "",
            "param": "",
            "value": ""
        }
    ],
    "extension_domain": [
        {
            "extension_schema": "http://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": "O-GlcNAc Glycosylation Sites (O-GlcNAc Atlas)",
                        "category_name": "tags"
                    },
                    {
                        "category_value": "csv",
                        "category_name": "file_type"
                    },
                    {
                        "category_value": "Reviewed",
                        "category_name": "status"
                    },
                    {
                        "category_value": "internal",
                        "category_name": "scope"
                    }
                ]
            }
        }
    ]
}
