{
    "io_domain": {
        "input_subdomain": [
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2wwwdata/reviewed/human_protein_masterlist.csv", 
                    "filename": "human_protein_masterlist.csv"
                }
            }, 
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2wwwdata/reviewed/mouse_protein_masterlist.csv", 
                    "filename": "rat_protein_masterlist.csv"
                }
            }, 
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2wwwdata/reviewed/rat_protein_masterlist.csv", 
                    "filename": "rat_protein_masterlist.csv"
                }
            }
        ], 
        "output_subdomain": [
            {
                "mediatype": "tsv", 
                "uri": {
                    "uri": "https://data.glygen.org/ln2wwwdata/reviewed/protein_glygen_uniprotkb_xref_mapping.tsv", 
                    "filename": "protein_glygen_uniprotkb_xref_mapping.tsv"
                }
            }, 
            {
                "mediatype": "csv", 
                "uri": {
                    "uri": "https://data.glygen.org/ln2wwwdata/reviewed/protein_glygen_uniprotkb_xref_mapping.stat.csv", 
                    "filename": "protein_glygen_uniprotkb_xref_mapping.stat.csv"
                }
            }
        ]
    }, 
    "spec_version": "1.3.0", 
    "object_id": "https://biocomputeobject.org/GLY_000410/1.5.30", 
    "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": "", 
                        "category_name": "species"
                    }, 
                    {
                        "category_value": "Protein", 
                        "category_name": "molecule"
                    }, 
                    {
                        "category_value": "GlyGen-UniProtKB Crossreferences", 
                        "category_name": "tags"
                    }, 
                    {
                        "category_value": "tsv", 
                        "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": "GlyGen-UniProtKB Crossreferences Mapping", 
        "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"
            }
        ], 
        "created": "2020-04-01T12:08:42.022917", 
        "modified": "2020-06-04T13:32:16.780973", 
        "version": "1.5.30"
    }, 
    "description_domain": {
        "keywords": [
            "UniProtKB", 
            "GlyGen", 
            "crossreferences", 
            "human", 
            "mouse", 
            "rat"
        ], 
        "pipeline_steps": [
            {
                "name": "make-species-agnostic-dataset.py", 
                "step_number": 1, 
                "version": "", 
                "input_list": [
                    {
                        "uri": "https://data.glygen.org/ln2wwwdata/reviewed/human_protein_masterlist.csv", 
                        "filename": "human_protein_masterlist.csv"
                    }, 
                    {
                        "uri": "https://data.glygen.org/ln2wwwdata/source/mouse_protein_masterlist.csv", 
                        "filename": "mouse_protein_masterlist.csv"
                    }, 
                    {
                        "uri": "https://data.glygen.org/ln2wwwdata/source/rat_protein_masterlist.csv", 
                        "filename": "rat_protein_masterlist.csv"
                    }
                ], 
                "output_list": [
                    {
                        "uri": "https://data.glygen.org/ln2wwwdata/reviewed/protein_glygen_uniprotkb_xref_mapping.tsv", 
                        "filename": "protein_glygen_uniprotkb_xref_mapping.tsv"
                    }
                ], 
                "prerequisite": [], 
                "description": "The script loads the GlyGen's UniProtKB protein accessions for human, mouse and rat species and extracts it into a csv dataset file"
            }
        ]
    }, 
    "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-species-agnostic-dataset.py", 
                    "filename": "make-species-agnostic-dataset.py"
                }
            }
        ]
    }, 
    "parametric_domain": [
        {
            "step": "", 
            "value": "", 
            "param": ""
        }
    ], 
    "usability_domain": [
        "The GlyGen-UniProtKB Crossreferences dataset contains UniProtKB glycoprotein accessions for human [taxid:9606], mouse [taxid:10090], rat [taxid:10116], hcv1a [taxid:11108], hcv1b [taxid:11116], sarscov1 [taxid:694009] and sarscov2 [taxid:2697049] glycoproteins for the purpose of UniProtKB to GlyGen crossreferences. In the dataset, the UniProtKB glycoprotein accessions have corresponding glycosylation annotation present in the datasets. The glycosylation annotation provides information about total (N,O,S,C) glycosylation sites followed by (if available) reported N-glycans and/or reported O-glycans with the number of sites. If you use this dataset please provide proper attribution to EMBL-EBI-UniProtKB, NCBI RefSeq and GlyGen"
    ], 
    "etag": "b546f69e618b5fe6e39ead05995dec87", 
    "_id": {
        "$oid": "5e84bc8af5c47b7eedc5aaf0"
    }
}