{
    "object_id": "https://biocomputeobject.org/GLY_001080/v-2.12.1",
    "spec_version": "1.3.0",
    "provenance_domain": {
        "name": "Arabidopsis Transcript Locus (Ensembl Transcript coordinates)",
        "version": "v-2.12.1",
        "license": "https://creativecommons.org/licenses/by/4.0/",
        "modified": "2025-11-04T12:27:54.952Z",
        "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": "Luke Johnson",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "luke.johnson@email.gwu.edu"
            },
            {
                "name": "Urnisha Bhuiyan",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "sbhuiyan28@email.gwu.edu"
            },
            {
                "name": "Cyrus Au-Yeung",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "cyruschauyeung@gwu.edu"
            },
            {
                "name": "Kate Warner",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "k.warner1@gwu.edu"
            }
        ]
    },
    "usability_domain": [
        "The Arabidopsis Transcript Locus (Ensembl Transcript coordinates) dataset contains UniProtKB Arabidopsis thaliana [taxid:3702] canonical accessions mapped to their isoforms with further mapping to the corresponding ENSEMBL Transcript ID, Peptide ID, chromosome ID along with the transcripts' start and end position. If you use this dataset please provide proper attribution to EMBL-EBI-UniProtKB, Ensembl and GlyGen. "
    ],
    "description_domain": {
        "keywords": [
            "transcript locus",
            "ensembl transcript id",
            "ensembl peptide id",
            "chromosome",
            "Ensembl"
        ],
        "pipeline_steps": [
            {
                "name": "make-protein-dataset.py",
                "step_number": 1,
                "input_list": [],
                "description": "The script loads the UniProtKB accessions that are mapped to ENSEMBL Transcript and Peptide identifiers from the triplestore containing triples from nt file, and then maps any UniProtKB isoforms to the UniProtKB canonical accession. The transcript ID is then used to extract the transcript's chromosome, start position, end position and forward/reverse strand data from the Ensembl gene annotation GFF3 file which has been converted to a TSV file. The extracted data is then used to create the csv dataset. ",
                "prerequisite": [],
                "output_list": []
            }
        ]
    },
    "parametric_domain": [
        {
            "value": ""
        }
    ],
    "io_domain": {
        "input_subdomain": [
            {
                "uri": {
                    "filename": "arabidopsis_protein_masterlist.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.12.1/reviewed/arabidopsis_protein_masterlist.csv"
                }
            },
            {
                "uri": {
                    "filename": "arabidopsis_ensembl_coords.tsv",
                    "uri": "https:/data.glygen.org/ln2downloads/ensembl/2026_06_01/arabidopsis_ensembl_coords.tsv"
                }
            },
            {
                "uri": {
                    "filename": "Arabidopsis_thaliana.TAIR10.63.uniprot.tsv",
                    "uri": "https:/data.glygen.org/ln2downloads/ensembl/2026_06_01/Arabidopsis_thaliana.TAIR10.63.uniprot.tsv"
                }
            }
        ],
        "output_subdomain": [
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.12.1/reviewed/arabidopsis_protein_transcriptlocus.csv",
                    "filename": "arabidopsis_protein_transcriptlocus.csv"
                }
            },
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.12.1/reviewed/arabidopsis_protein_transcriptlocus.stat.csv",
                    "filename": "arabidopsis_protein_transcriptlocus.stat.csv"
                }
            }
        ]
    },
    "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": "Arabidopsis thaliana",
                        "category_name": "species"
                    },
                    {
                        "category_value": "Protein",
                        "category_name": "molecule"
                    },
                    {
                        "category_value": "Transcript Coordinates",
                        "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": "locus"
                    }
                ]
            }
        }
    ],
    "error_domain": {},
    "etag": "1ae23d43827da29a39edfb379936036d"
}