{
    "io_domain": {
        "input_subdomain": [
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2wwwdata/source/mutid-map.final.csv", 
                    "filename": "mutid-map.final.csv"
                }
            }, 
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2wwwdata/reviewed/human_protein_masterlist.csv", 
                    "filename": "human_protein_masterlist.csv"
                }
            }
        ], 
        "output_subdomain": [
            {
                "mediatype": "csv", 
                "uri": {
                    "uri": "https://data.glygen.org/ln2wwwdata/reviewed/human_protein_mutation.csv", 
                    "filename": "human_protein_mutation.csv"
                }
            }, 
            {
                "mediatype": "csv", 
                "uri": {
                    "uri": "https://data.glygen.org/ln2wwwdata/reviewed/human_protein_mutation.stat.csv", 
                    "filename": "human_protein_mutation.stat.csv"
                }
            }
        ]
    }, 
    "spec_version": "1.3.0", 
    "object_id": "https://biocomputeobject.org/GLY_000024/1.5.12", 
    "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": "Homo Sapiens", 
                        "category_name": "species"
                    }, 
                    {
                        "category_value": "Protein", 
                        "category_name": "molecule"
                    }, 
                    {
                        "category_value": "Somatic Cancer Mutations", 
                        "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": "Human Cancer Mutation", 
        "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": "BioMuta - The George Washington University", 
                "contribution": [
                    "importedFrom"
                ], 
                "name": "Robel Kahsay", 
                "email": "rykahsay@gwu.edu"
            }
        ], 
        "created": "2019-01-23T17:21:05.891293", 
        "modified": "2019-11-28T10:58:39.643598", 
        "version": "1.5.12"
    }, 
    "description_domain": {
        "keywords": [
            "single-nucleotide variations (nsSNVs)", 
            "cancer mutations", 
            "BioMuta", 
            "disease"
        ], 
        "pipeline_steps": [
            {
                "name": "make-protein-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/mutid-map.final.csv", 
                        "filename": "mutid-map.final.csv"
                    }
                ], 
                "output_list": [
                    {
                        "uri": "https://data.glygen.org/ln2wwwdata/reviewed/human_protein_mutation.csv", 
                        "filename": "human_protein_mutation.csv"
                    }
                ], 
                "prerequisite": [], 
                "description": "The script extracts the mutation data that has cancer mutation frequency of greater than of 0.01(1%) and maps the UniProtKB accession to the UniProtKB canonical accessions"
            }
        ]
    }, 
    "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"
                }
            }
        ]
    }, 
    "error_domain": {
        "empirical_error": {}, 
        "algorithmic_error": {}, 
        "emperical_error": {}
    }, 
    "parametric_domain": [
        {
            "step": "", 
            "value": "", 
            "param": ""
        }
    ], 
    "usability_domain": [
        "The Human Cancer Mutation dataset contains human [taxid:9606] mutation data from BioMuta database (https://hive.biochemistry.gwu.edu/biomuta). The current version of BioMuta contains only nonsynonymous single-nucleotide variations (nsSNVs) associated with cancer. BioMuta is a single-nucleotide variation (SNV) and disease association database where variations are mapped to genomes and NCBI RefSeq nucleotide entries, and unified through UniProtKB/Swiss-Prot positional coordinates. The dataset is imported and contributed by BioMuta and processed by GlyGen. If you use this dataset please provide proper attribution to BioMuta and GlyGen. Note GlyGen only integrates and displays the data that has cancer mutation frequency of greater than of 0.01(1%)"
    ], 
    "etag": "b3c0990e45f4c00e7f0d06f0975e7737", 
    "_id": {
        "$oid": "5c48e8d1f5c47b65ae4b5d03"
    }
}