{
    "object_id": "https://biocomputeobject.org/GLY_000030/1.0.13", 
    "license": "https://creativecommons.org/licenses/by/4.0/", 
    "contributors": [
        {
            "orcid": "", 
            "affiliation": "", 
            "contribution": [
                "createdBy"
            ], 
            "name": "Jeet Vora", 
            "email": "jeetvora@gwu.edu"
        }, 
        {
            "orcid": "", 
            "affiliation": "", 
            "contribution": [
                "createdBy"
            ], 
            "name": "Rahi Navelkar", 
            "email": "rsn13@gwu.edu"
        }
    ], 
    "created": "2019-01-23T17:21:05.897519", 
    "spec_version": "1.3.0", 
    "modified": "2019-02-21T11:52:15.668750", 
    "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": "Reviewed", 
                        "category_name": "status"
                    }
                ]
            }
        }
    ], 
    "io_domain": {
        "input_subdomain": [
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2wwwdata/source/mouse_protein_glycosyltransferase_2018_11.csv", 
                    "filename": "mouse_protein_glycosyltransferase_2018_11.csv"
                }
            }, 
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2wwwdata/reviewed/protein_ortholog.csv", 
                    "filename": "protein_ortholog.csv"
                }
            }, 
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2wwwdata/reviewed/human_protein_idmapping.csv", 
                    "filename": "human_protein_idmapping.csv"
                }
            }
        ], 
        "output_subdomain": [
            {
                "mediatype": "csv", 
                "uri": {
                    "uri": "https://data.glygen.org/ln2wwwdata/reviewed/mouse_protein_glycosyltransferase.csv", 
                    "filename": "mouse_protein_glycosyltransferase.csv"
                }
            }, 
            {
                "mediatype": "csv", 
                "uri": {
                    "uri": "https://data.glygen.org/ln2wwwdata/reviewed/mouse_protein_glycosyltransferase.stat.csv", 
                    "filename": "mouse_protein_glycosyltransferase.stat.csv"
                }
            }
        ]
    }, 
    "provenance_domain": {
        "license": "https://creativecommons.org/licenses/by/4.0/", 
        "name": "Mouse Glycosyltransferase", 
        "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": "2019-01-23T17:21:05.897519", 
        "modified": "2019-04-23T14:40:45.888291", 
        "version": "1.0.13"
    }, 
    "description_domain": {
        "keywords": [
            "GT", 
            "glycosyltransferase", 
            "CAZy", 
            "CFG", 
            "GO"
        ], 
        "pipeline_steps": [
            {
                "name": "extract-from-csv.py", 
                "step_number": 1, 
                "version": "1.0", 
                "input_list": [
                    {
                        "uri": "https://data.glygen.org/ln2wwwdata/source/mouse_protein_glycosyltransferase_2018_11.csv", 
                        "filename": "mouse_protein_glycosyltransferase_2018_11.csv"
                    }, 
                    {
                        "uri": "https://data.glygen.org/ln2wwwdata/reviewed/protein_ortholog.csv", 
                        "filename": "protein_ortholog.csv"
                    }, 
                    {
                        "uri": "https://data.glygen.org/ln2wwwdata/reviewed/mouse_protein_idmapping.csv", 
                        "filename": "mouse_protein_idmapping.csv"
                    }
                ], 
                "output_list": [
                    {
                        "uri": "https://data.glygen.org/ln2wwwdata/reviewed/mouse_protein_glycosyltransferase.csv", 
                        "filename": "mouse_protein_glycosyltransferase.csv"
                    }
                ], 
                "prerequisite": [], 
                "description": "The script extract-from-csv.py extracts the columns from source file that contains the list of GTs and maps the UniProtKB accessions to current UniprotKB canonical accessions. Creating the source file - The mouse orthologs for human glycosyltransferases were retrieved from protein_ortholog.csv https://data.glygen.org/ln2wwwdata/reviewed/protein_ortholog.csv using the vlookup function in the Excel spreadsheet.  The retrieved mouse UniProtKB accessions were searched in Quick GO, CAZY database, Brenda for evidence of glycosyltransferase activity. The accessions that did not have requisite evidence were filtered out. The UniProt accessions for keyword glycosyltransferase[UniProt KW-0328] were searched in UniProt but that resulted in fewer accessions than above step."
            }
        ]
    }, 
    "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/pipeline/extractor/extract-from-csv.py", 
                    "filename": "extract-from-csv.py"
                }
            }
        ]
    }, 
    "version": "1.0.5", 
    "error_domain": {
        "empirical_error": {}, 
        "algorithmic_error": {}, 
        "emperical_error": {}
    }, 
    "parametric_domain": [
        {
            "step": "", 
            "value": "", 
            "param": ""
        }
    ], 
    "usability_domain": [
        "The Mouse Glycosyltransferase dataset contains list of mouse [taxid:10090] glycosyltransferases enzymes", 
        "If you use this dataset please provide proper attribution to GlyGen"
    ], 
    "etag": "8072d59338ac9c2e3466aad22df0d2c0", 
    "_id": {
        "$oid": "5c48e8d1f5c47b65ae4b5d09"
    }, 
    "name": "Mouse Glycosyltransferase"
}