{
    "io_domain": {
        "input_subdomain": [
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2data/releases/data/v-1.5.18/source/dbSNP-homo-sapiens.tsv",
                    "filename": "dbSNP-homo-sapiens.tsv"
                }
            },
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2data/releases/data/v-1.5.18/reviewed/human_protein_masterlist.csv",
                    "filename": "human_protein_masterlist.csv"
                }
            }
        ],
        "output_subdomain": [
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https://data.glygen.org/ln2data/releases/data/v-1.5.18/reviewed/human_protein_mutation_germline.csv",
                    "filename": "human_protein_mutation_germline.csv"
                }
            },
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https://data.glygen.org/ln2data/releases/data/v-1.5.18/reviewed/human_protein_mutation_germline.stat.csv",
                    "filename": "human_protein_mutation_germline.stat.csv"
                }
            }
        ]
    },
    "spec_version": "1.3.0",
    "object_id": "https://biocomputeobject.org/GLY_000459/v-1.5.18",
    "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": "Germline 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": "",
        "name": "Human Germline 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": "EMBL-EBI-UniProt",
                "contribution": [
                    "contributedBy"
                ],
                "name": "Maria Martin",
                "email": "martin@ebi.ac.uk"
            },
            {
                "affiliation": "EMBL-EBI-UniProt",
                "contribution": [
                    "contributedBy"
                ],
                "name": "Preethi Vasudev",
                "email": "pvasudev@ebi.ac.uk"
            }
        ],
        "created": "2020-04-03T17:17:42.572968",
        "modified": "2020-04-03T17:41:28.937559",
        "version": "v-1.5.18"
    },
    "description_domain": {
        "keywords": [
            "single-nucleotide variations (nsSNVs)",
            "germline mutations",
            "dbSNP",
            "disease",
            "ClinVar"
        ],
        "pipeline_steps": [
            {
                "name": "make-protein-dataset.py",
                "step_number": 1,
                "version": "",
                "input_list": [
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-1.5.18/reviewed/human_protein_masterlist.csv",
                        "filename": "human_protein_masterlist.csv"
                    },
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-1.5.18/source/dbSNP-homo-sapiens.tsv",
                        "filename": "dbSNP-homo-sapiens.tsv"
                    }
                ],
                "output_list": [
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-1.5.18/reviewed/human_protein_mutation_cancer.csv",
                        "filename": "human_protein_mutation_cancer.csv"
                    }
                ],
                "prerequisite": [],
                "description": "The script extracts the gernline (somatic_staus = 0) mutation data from the input file 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 Germline Mutation dataset contains human [taxid:9606] germline mutation data from EBI Variant Data API. Only nonsynonymous single-nucleotide variations (nsSNVs) germline mutations are integrated. The dataset is from 2019-11 UniProtKB release .If you use this dataset provide proper attribution to GlyGen and EBI-EMBL-UniProtKB."
    ],
    "etag": "d00ff0fe4695f6bc0c49ad32e6729c71",
    "_id": {
        "$oid": "5e87a7f6f5c47b0e62d111ef"
    }
}
