{
    "object_id": "https://biocomputeobject.org/GLY_001687/v-2.12.1",
    "etag": "182179f26288306c99323ec3b056e64d",
    "spec_version": "1.3.0",
    "provenance_domain": {
        "name": "Arabidopsis Domains and Regions Citations (UniProtKB)",
        "version": "v-2.12.1",
        "license": "https://creativecommons.org/licenses/by/4.0/",
        "created": "",
        "modified": "",
        "contributors": [
            {
                "name": "Jeet Vora",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "jeetvora@gwu.edu"
            },
            {
                "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 domain and regions citations (UniProtKB) dataset contains Arabidopsis thaliana [taxid:3702] publication information for UniProtKB canonical protein domains, motifs and DNA binding regions provided in GlyGen. The dataset includes PMID, journal name, date and author and title. If you use this dataset please give proper attribution to UniProtKB and GlyGen."
    ],
    "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": "Domains and Regions Citations",
                        "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": "citations"
                    }
                ]
            }
        }
    ],
    "description_domain": {
        "keywords": [
            "domains",
            "motifs",
            "regions",
            "DNA binding",
            "citations"
        ],
        "pipeline_steps": [
            {
                "name": "make-protein-dataset.py",
                "step_number": 1,
                "input_list": [],
                "description": "The script extracts the PMID from the domain, motif or nucleotide binding annotation triples then accesses the NCBI ENTREZ infrastructure to retrieve the publication information of the given PMID from UniProtKB.",
                "prerequisite": [],
                "output_list": []
            }
        ]
    },
    "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/glygener/glygen-backend-integration/blob/master/dataset-maker/make-protein-dataset.py",
                    "filename": "make-protein-dataset.py"
                }
            }
        ]
    },
    "io_domain": {
        "output_subdomain": [
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.12.1/reviewed/arabidopsis_protein_citations_domains_uniprotkb.csv",
                    "filename": "arabidopsis_protein_citations_domains_uniprotkb.csv"
                }
            },
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.12.1/reviewed/arabidopsis_protein_citations_domains_uniprotkb.stat.csv",
                    "filename": "arabidopsis_protein_citations_domains_uniprotkb.stat.csv"
                }
            }
        ],
        "input_subdomain": [
            {
                "uri": {
                    "filename": "arabidopsis_protein_domains_uniprotkb.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.12.1/reviewed/arabidopsis_protein_domains_uniprotkb.csv"
                }
            }
        ]
    },
    "error_domain": {}
}