{
    "object_id": "https://biocomputeobject.org/GLY_000524/v-1.7.13",
    "spec_version": "1.3",
    "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": "Glycation Sites (UniProtKB)",
                        "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": "",
        "name": "Rat Glycation Sites (UniProtKB)",
        "contributors": [
            {
                "affiliation": "",
                "contribution": [
                    "createdBy"
                ],
                "name": "",
                "email": "jeetvora@gwu.edu"
            }
        ],
        "created": "2020-08-17T17:48:42.126261",
        "modified": "",
        "version": "v-1.7.13"
    },
    "description_domain": {
        "keywords": [],
        "pipeline_steps": [
            {
                "name": "make-proteoform-dataset.py",
                "step_number": 1,
                "version": "",
                "input_list": [
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-1.7.13/reviewed/rat_protein_masterlist.csv",
                        "filename": "rat_protein_masterlist.csv"
                    },
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-1.7.13/source/uniprot-proteome-rattus-norvegicus.nt",
                        "filename": "uniprot-proteome-rattus-norvegicus.nt"
                    }
                ],
                "output_list": [
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-1.7.13/reviewed/rat_proteoform_glycation_sites_uniprotkb.csv",
                        "filename": "rat_proteoform_glycation_sites_uniprotkb.csv"
                    }
                ],
                "prerequisite": [],
                "description": "The script loads the glycation information from the triplestore conatining triples from the nt file and extracts 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-proteoform-dataset.py",
                    "filename": "make-proteoform-dataset.py"
                }
            }
        ]
    },
    "usability_domain": [
        "The Rat Glycation Sites (UniProtKB) dataset contains mouse [taxid:10116] UniProtKB canonical accessions with information on reported and predicted glycation sites in UniProtKB database. The dataset is derived from 2019-09 UniProtKB release. If you use this dataset please provide proper attribution to EMBL-EBI-UniProtKB and GlyGen"
    ],
    "etag": "7d5e9cd99df4cc9e92a6a0d0e184f7c4",
    "_id": {
        "$oid": "5f3afb3af5c47b5ca32292d8"
    },
    "io_domain": {
        "input_subdomain": [
            {
                "uri": {
                    "uri": "https://data.glygen.org//ln2data/releases/data/v-1.7.13/reviewed/rat_protein_masterlist.csv",
                    "filename": "rat_protein_masterlist.csv"
                }
            },
            {
                "uri": {
                    "uri": "https://data.glygen.org//ln2downloads/ebi/2020_05/uniprot-proteome-rattus-norvegicus.nt",
                    "filename": "uniprot-proteome-rattus-norvegicus.nt"
                }
            }
        ],
        "output_subdomain": [
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "http://data.glygen.org/ln2data/releases/data/v-1.7.13/reviewed/rat_proteoform_glycation_sites_uniprotkb.csv",
                    "filename": "rat_proteoform_glycation_sites_uniprotkb.csv"
                }
            },
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "http://data.glygen.org/ln2data/releases/data/v-1.7.13/reviewed/rat_proteoform_glycation_sites_uniprotkb.stat.csv",
                    "filename": "rat_proteoform_glycation_sites_uniprotkb.stat.csv"
                }
            }
        ]
    }
}
