{
    "object_id": "https://biocomputeobject.org/GLY_000237/v-1.7.13",
    "spec_version": "1.3.0",
    "extension_domain": [
        {
            "extension_schema": "http://www.w3id.org/biocompute/extension_domain/1.1.0/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": "Rattus norvegicus",
                        "category_name": "species"
                    },
                    {
                        "category_value": "Protein",
                        "category_name": "molecule"
                    },
                    {
                        "category_value": "Transcript Coordinates",
                        "category_name": "tags"
                    },
                    {
                        "category_value": "csv",
                        "category_name": "file_type"
                    },
                    {
                        "category_value": "Reviewed",
                        "category_name": "status"
                    },
                    {
                        "category_value": "internal",
                        "category_name": "scope"
                    }
                ]
            }
        }
    ],
    "provenance_domain": {
        "license": "https://creativecommons.org/licenses/by/4.0/",
        "name": "Rat Transcript Locus (Ensembl Transcript coordinates)",
        "contributors": [
            {
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "name": "Jeet Vora",
                "email": "jeetvora@gwu.edu"
            },
            {
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "name": "Rahi Navelkar",
                "email": "rsn13@gwu.edu"
            },
            {
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "name": "Robel Kahsay",
                "email": "rykahsay@gwu.edu"
            },
            {
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "name": "Xavier Holmes",
                "email": "xavierh@email.gwu.edu"
            },
            {
                "affiliation": "EMBL-EBI-UniProt",
                "contribution": [
                    "contributedBy"
                ],
                "name": "Maria Martin",
                "email": "martin@ebi.ac.uk"
            },
            {
                "affiliation": "EMBL-EBI-UniProt",
                "contribution": [
                    "contributedBy"
                ],
                "name": "Preethi Vasudev",
                "email": "pvasudev@ebi.ac.uk"
            }
        ],
        "created": "2019-09-06T14:06:38.964793",
        "modified": "2020-04-10T11:55:37.315722",
        "version": "v-1.7.13"
    },
    "description_domain": {
        "keywords": [
            "transcript locus",
            "ensembl transcript id",
            "ensembl peptide id",
            "chromosome",
            "Ensembl"
        ],
        "pipeline_steps": [
            {
                "name": "make-protein-dataset.py",
                "step_number": 1,
                "version": "",
                "input_list": [
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-1.7.13/reviewed/rat_protein_idmapping.csv",
                        "filename": "rat_protein_idmapping.csv"
                    },
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-1.7.13/source/uniprot-proteome-10116-rattus-norvegicus-2019-05.nt",
                        "filename": "uniprot-proteome-10116-rattus-norvegicus-2019-05.nt"
                    }
                ],
                "output_list": [
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-1.7.13/reviewed/rat_protein_transcriptlocus.csv",
                        "filename": "rat_protein_transcriptlocus.csv"
                    }
                ],
                "prerequisite": [],
                "description": "The script loads the UniProtKB accessions that are mapped to UniProtKB isoforms and ENSEMBL Transcript and Peptide identifiers from the triplestore containing triples from nt file and maps it to the UniProtKB canonical accession to finally extract the data as a csv dataset"
            }
        ]
    },
    "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/GW-HIVE/glygen-backend-integration/blob/master/pipeline/integrator/make-protein-dataset.py",
                    "filename": "make-protein-dataset.py"
                }
            }
        ]
    },
    "usability_domain": [
        "The Rat Ensembl Transcript Locus (Transcript coordinates)s contains UniProtKB rat [taxid:10116] canonical accessions mapped to their isoforms with further mapping to the corresponding ENSEMBL Transcript ID, Peptide ID, chromosome ID along with the transcripts's start and end position The dataset is derived from 2019-05 UniProt release. If you use this dataset please provide proper attribution to EMBL-EBI-UniProtKB and GlyGen"
    ],
    "etag": "f245e6f3beaabf946882d458122766e9",
    "_id": {
        "$oid": "5d72a02ef5c47b0f4342335e"
    },
    "io_domain": {
        "input_subdomain": [
            {
                "uri": {
                    "uri": "https://data.glygen.org//ln2downloads/ebi/2020_05/uniprot-proteome-rattus-norvegicus.nt",
                    "filename": "uniprot-proteome-10116-rattus-norvegicus-2019-05.nt"
                }
            },
            {
                "uri": {
                    "uri": "https://data.glygen.org//ln2data/releases/data/v-1.7.13/reviewed/rat_protein_masterlist.csv",
                    "filename": "rat_protein_idmapping.csv"
                }
            }
        ],
        "output_subdomain": [
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "http://data.glygen.org/ln2data/releases/data/v-1.7.13/reviewed/rat_protein_transcriptlocus.csv",
                    "filename": "rat_protein_transcriptlocus.csv"
                }
            },
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "http://data.glygen.org/ln2data/releases/data/v-1.7.13/reviewed/rat_protein_transcriptlocus.stat.csv",
                    "filename": "rat_protein_transcriptlocus.stat.csv"
                }
            }
        ]
    }
}