{
    "io_domain": {
        "input_subdomain": [
            {
                "uri": {
                    "access_time": "", 
                    "sha1_chksum": "", 
                    "uri": "https://data.glygen.org//ln2releases/data/v-1.5.36/reviewed/mouse_protein_masterlist.csv", 
                    "filename": "mouse_protein_masterlist.csv"
                }
            }, 
            {
                "uri": {
                    "access_time": "", 
                    "sha1_chksum": "", 
                    "uri": "https://data.glygen.org//ln2downloads/matrixdb/2019_08_07/matrixdb_CORE.csv", 
                    "filename": "matrixdb_CORE.csv"
                }
            }
        ], 
        "output_subdomain": [
            {
                "mediatype": "csv", 
                "uri": {
                    "access_time": "2020-02-06T13:47:43-0500", 
                    "sha1_chksum": "", 
                    "uri": "http://data.glygen.org/ln2wwwdata/reviewed/mouse_protein_matrixdb.csv", 
                    "filename": "mouse_protein_matrixdb.csv"
                }
            }, 
            {
                "mediatype": "csv", 
                "uri": {
                    "access_time": "", 
                    "sha1_chksum": "", 
                    "uri": "http://data.glygen.org/ln2wwwdata/reviewed/mouse_protein_matrixdb.stat.csv", 
                    "filename": "mouse_protein_matrixdb.stat.csv"
                }
            }
        ]
    }, 
    "bco_spec_version": "1.3.0", 
    "checksum": "c964f1fb351853e6b006ac5f0bd6f7e2", 
    "extension_domain": {
        "additional_license": {
            "data_license": "https://creativecommons.org/licenses/by/4.0/", 
            "script_license": "https://www.gnu.org/licenses/gpl-3.0.en.html"
        }, 
        "fhir_extension": [], 
        "scm_extension": {
            "scm_repository": "", 
            "scm_path": "", 
            "scm_type": "git", 
            "scm_commit": ""
        }, 
        "dataset_categories": [
            {
                "category_value": "Mus musculus", 
                "category_name": "species"
            }, 
            {
                "category_value": "Protein", 
                "category_name": "molecule"
            }, 
            {
                "category_value": "MatrixDB", 
                "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": "Mouse Interaction data (MatrixDB)", 
        "contributors": [
            {
                "orcid": "", 
                "affiliation": "The George Washington University", 
                "contribution": [
                    "createdBy"
                ], 
                "name": "Rahi Navelkar", 
                "email": "rsn13@gwu.edu"
            }, 
            {
                "orcid": "", 
                "affiliation": "The George Washington University", 
                "contribution": [
                    "createdBy"
                ], 
                "name": "Jeet Vora", 
                "email": "jeetvora@gwu.edu"
            }, 
            {
                "orcid": "", 
                "affiliation": "The George Washington University", 
                "contribution": [
                    "createdBy"
                ], 
                "name": "Robel Kahsay", 
                "email": "rykahsay@gwu.edu"
            }, 
            {
                "orcid": "", 
                "affiliation": "MatrixDB", 
                "contribution": [
                    "contributedBy"
                ], 
                "name": "Sylvie Ricard-Blum", 
                "email": "sylvie.ricard-blum@univ-lyon1.fr"
            }
        ], 
        "created": "2020-02-10T14:02:01.224476", 
        "modified": "2020-05-01T15:23:29.016841", 
        "version": "1.5.36"
    }, 
    "description_domain": {
        "keywords": [
            "protein-GAGs interaction", 
            "MatrixDB"
        ], 
        "pipeline_steps": [
            {
                "name": "make-protein-dataset.py", 
                "step_number": 1, 
                "version": "", 
                "input_list": [
                    {
                        "access_time": "2020-02-06T13:45:43-0500", 
                        "uri": "https://data.glygen.org/ln2wwwdata/source/matrixdb_CORE.csv", 
                        "filename": "matrixdb_CORE.csv"
                    }, 
                    {
                        "access_time": "2020-02-06T13:45:43-0500", 
                        "uri": "https://data.glygen.org/ln2wwwdata/reviewed/mouse_protein_masterlist.csv", 
                        "filename": "mouse_protein_masterlist.csv"
                    }
                ], 
                "output_list": [
                    {
                        "access_time": "2020-02-06T13:47:43-0500", 
                        "uri": "https://data.glygen.org/ln2wwwdata/reviewed/mouse_protein_matrixdb.csv", 
                        "filename": "mouse_protein_matrixdb.csv"
                    }
                ], 
                "prerequisite": [], 
                "description": "The tab file (matrixdb_CORE.tab (labeled as \u201cCore MatrixDB dataset as PSI-MI TAB 2.7 file\u201d) was downloaded directly from the MatrixDB database. Following rules were applied to extract specific data from the file. The rules were also verified by the author. 1)Rule 1: where column \"Alt. ID(s) interactor A\" has \"GAG\" AND column \"ID(s) interactor B\" has \"uniprotkb\" as prefixs. 2)Rule 2: where column \"Alt. ID(s) interactor B\" has \"GAG\" AND column \"#ID(s) interactor A\" has \"uniprotkb\" as prefixes. 3)for Rule 1: 3a) Add values from column \"Biological role(s) interactor A\" to new column \"gag_interactor_biological_role\". 3b)Add from column \"Biological role(s) interactor B\" to new column \"protein_interactor_biological_role\".3c)Add  from column \"Publication Identifier(s)\" to new column \"evidence\". 4)for Rule 2: 4a)Add values from\"Biological role(s) interactor A\" to column \"protein_interactor_biological_role\". 4b)Add  from \"Biological role(s) interactor B\" to column gag_interactor_biological_role\". 4c)Add from \"Publication Identifier(s)\" to column \"evidence\". 5) finally the UniProtKB accessions were mapped to UniProtKB canonical accessions based on the GlyGen protein master list (https://data.glygen.org/GLYDS000001)."
            }
        ]
    }, 
    "execution_domain": {
        "external_data_endpoints": [], 
        "environment_variables": {}, 
        "script_driver": "manual", 
        "software_prerequisites": [
            {
                "version": "2.7.5", 
                "name": "Python", 
                "uri": {
                    "access_time": "2019-01-07T17:43:17-0500", 
                    "sha1_chksum": "", 
                    "uri": "https://www.python.org/download/releases/2.7.5/"
                }
            }
        ], 
        "script": [
            {
                "uri": {
                    "access_time": "2019-07-26T17:43:17-0500", 
                    "sha1_chksum": "", 
                    "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": {}
    }, 
    "parametric_domain": [], 
    "usability_domain": [
        "The Mouse Interaction data (MatrixDB) dataset contains interaction data between Mouse [taxid:10090] proteins and Glycosaminoglycans (GAGs) from MatrixDB database (http://matrixdb.univ-lyon1.fr/ ;PMID:19147664). If you use this dataset please provide proper attribution to MatrixDB and GlyGen."
    ], 
    "_id": {
        "$oid": "5e41a8a9f5c47b1551229d85"
    }, 
    "bco_id": "http://data.glygen.org/GLY_000333"
}