{
    "object_id": "https://biocomputeobject.org/GLY_001248/v-2.9.1",
    "spec_version": "1.3.0",
    "etag": "8fdba489078f2ff53d98adb76551bfbe",
    "provenance_domain": {
        "name": "Arabidopsis UniProtKB Xref Araport",
        "version": "v-2.9.1",
        "license": "https://creativecommons.org/licenses/by/4.0/",
        "modified": "2024-11-11T20:28:07.435Z",
        "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 Arabidopsis UniprotKB Xref Arabidopsis Information Portal (Araport) contains Arabidopsis thaliana [taxid:3702] UniProtKB canonical accessions cross-referenced to Araport database accessions/identifiers. If you use this dataset please provide proper attribution to EMBL-EBI-UniProtKB and GlyGen.",
        "Araport was a database of Arabidopsis gene sequences and associated annotations that has been discontinued. Teams from The Arabidopsis Information Resource (TAIR), the National Center for Genome Resources (NCGR), and the Bio-Analytic Resource for Plant Biology (BAR) have taken over its operation and have refreshed/expanded the functionalities that were available at Araport."
    ],
    "description_domain": {
        "pipeline_steps": [
            {
                "name": "make-protein-dataset.py",
                "step_number": 1,
                "input_list": [],
                "description": "The script loads the UniProtKB canonical accessions that are cross-referenced to Araport database accessions/identifiers from the triplestore containing triples from nt file and extracts the data as a csv dataset",
                "prerequisite": [],
                "output_list": []
            }
        ],
        "keywords": [
            "Araport",
            "cross-reference"
        ]
    },
    "parametric_domain": [
        {
            "value": ""
        }
    ],
    "io_domain": {
        "output_subdomain": [
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.9.1/reviewed/arabidopsis_protein_xref_araport.csv",
                    "filename": "arabidopsis_protein_xref_araport.csv"
                }
            },
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.9.1/reviewed/arabidopsis_protein_xref_araport.stat.csv",
                    "filename": "arabidopsis_protein_xref_araport.stat.csv"
                }
            }
        ],
        "input_subdomain": [
            {
                "uri": {
                    "filename": "uniprot-proteome-arabidopsis-thaliana.nt",
                    "uri": "https:/data.glygen.org/ln2downloads/ebi/2025_05_20/uniprot-proteome-arabidopsis-thaliana.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": [
        {
            "dataset_extension": {
                "dataset_categories": [
                    {
                        "category_value": "Arabidopsis thaliana",
                        "category_name": "species"
                    },
                    {
                        "category_value": "Protein",
                        "category_name": "molecule"
                    },
                    {
                        "category_value": "Araport cross-references",
                        "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"
        }
    ],
    "error_domain": {}
}