{
    "io_domain": {
        "input_subdomain": [
            {
                "uri": {
                    "access_time": "", 
                    "sha1_chksum": "", 
                    "uri": "https://data.glygen.org//ln2downloads/ncbi/refseq/2020_12/refseq_protein_all_9606.gpff", 
                    "filename": "refseq_protein_all_9606.gpff"
                }
            }, 
            {
                "uri": {
                    "access_time": "", 
                    "sha1_chksum": "", 
                    "uri": "https://data.glygen.org//ln2releases/data/v-1.11.2/reviewed/human_protein_xref_refseq.csv", 
                    "filename": "human_protein_xref_refseq.csv"
                }
            }
        ], 
        "output_subdomain": [
            {
                "mediatype": "csv", 
                "uri": {
                    "access_time": "2019-10-26T17:45:17-0500", 
                    "sha1_chksum": "", 
                    "uri": "http://data.glygen.org/ln2wwwdata/reviewed/human_protein_function_refseq.csv", 
                    "filename": "human_protein_function_refseq.csv"
                }
            }, 
            {
                "mediatype": "csv", 
                "uri": {
                    "access_time": "2019-10-26T17:45:17-0500", 
                    "sha1_chksum": "", 
                    "uri": "http://data.glygen.org/ln2wwwdata/reviewed/human_protein_function_refseq.stat.csv", 
                    "filename": "human_protein_function_refseq.stat.csv"
                }
            }
        ]
    }, 
    "bco_spec_version": "1.3.0", 
    "checksum": "2d5000b0fcd78b7d155ec060f71be911", 
    "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": "Homo sapiens", 
                "category_name": "species"
            }, 
            {
                "category_value": "Protein", 
                "category_name": "molecule"
            }, 
            {
                "category_value": "Protein Function", 
                "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 Protein Function (NCBI Refseq)", 
        "contributors": [
            {
                "orcid": "", 
                "affiliation": "The George Washington University", 
                "contribution": [
                    "createdBy"
                ], 
                "name": "Jeet Vora", 
                "email": "jeetvora@gwu.edu"
            }, 
            {
                "orcid": "", 
                "affiliation": "The George Washington University", 
                "contribution": [
                    "createdBy"
                ], 
                "name": "Rahi Navelkar", 
                "email": "rsn13@gwu.edu"
            }, 
            {
                "orcid": "", 
                "affiliation": "The George Washington University", 
                "contribution": [
                    "createdBy"
                ], 
                "name": "Robel Kahsay", 
                "email": "rykahsay@gwu.edu"
            }, 
            {
                "orcid": "", 
                "affiliation": "The George Washington University", 
                "contribution": [
                    "createdBy"
                ], 
                "name": "Xavier Holmes", 
                "email": "xavierh@email.gwu.edu"
            }, 
            {
                "orcid": "", 
                "affiliation": "EMBL-EBI-UniProt", 
                "contribution": [
                    "contributedBy"
                ], 
                "name": "Maria Martin", 
                "email": "martin@ebi.ac.uk"
            }, 
            {
                "orcid": "", 
                "affiliation": "EMBL-EBI-UniProt", 
                "contribution": [
                    "contributedBy"
                ], 
                "name": "Preethi Vasudev", 
                "email": "pvasudev@ebi.ac.uk"
            }, 
            {
                "orcid": "", 
                "affiliation": "The George Washington University", 
                "contribution": [
                    "createdBy"
                ], 
                "name": "Karina Martinez", 
                "email": "karinamartinez@gwu.edu"
            }, 
            {
                "orcid": "", 
                "affiliation": "The George Washington University", 
                "contribution": [
                    "createdBy"
                ], 
                "name": "Amanda Bell", 
                "email": "amandab2140@gwu.edu"
            }, 
            {
                "orcid": "", 
                "affiliation": "The George Washington University", 
                "contribution": [
                    "createdBy"
                ], 
                "name": "Jonathon Keeney", 
                "email": "keeneyjg@gwu.edu"
            }, 
            {
                "orcid": "", 
                "affiliation": "The George Washington University", 
                "contribution": [
                    "createdBy"
                ], 
                "name": "Millicent Quartey", 
                "email": "mquartey@email.gwu.edu"
            }
        ], 
        "created": "2019-03-16T15:35:15.248632", 
        "modified": "2022-01-19T11:15:01.844413", 
        "version": "1.11.2"
    }, 
    "description_domain": {
        "keywords": [
            "protein function", 
            "NCBI RefSeq", 
            "geneRIF", 
            "PMID"
        ], 
        "pipeline_steps": [
            {
                "name": "make-protein-dataset.py", 
                "step_number": 1, 
                "version": "", 
                "input_list": [
                    {
                        "access_time": "2019-10-26T17:43:17-0500", 
                        "uri": "https://data.glygen.org/ln2wwwdata/reviewed/human_protein_xref_refseq.csv", 
                        "filename": "human_protein_xref_refseq.csv"
                    }, 
                    {
                        "access_time": "2019-10-26T17:43:17-0500", 
                        "uri": "https://data.glygen.org/ln2wwwdata/source/refseq_protein_all_9606.gpff", 
                        "filename": "refseq_protein_all_9606.gpff"
                    }
                ], 
                "output_list": [
                    {
                        "access_time": "2019-07-26T17:45:17-0500", 
                        "uri": "https://data.glygen.org/ln2wwwdata/reviewed/human_protein_info_refseq.csv", 
                        "filename": "human_protein_info_refseq.csv"
                    }
                ], 
                "prerequisite": [], 
                "description": "The script parses the NCBI RefSeq gpff files to extract the geneRIFs that contain protein function. The script also maps RefSeq 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": {
                    "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": {}, 
        "emperical_error": {}
    }, 
    "parametric_domain": [
        {
            "step": "", 
            "value": "", 
            "param": ""
        }
    ], 
    "usability_domain": [
        "The Human Protein Function (NCBI RefSeq) dataset contains biological function annotation of human [taxid:9606] genes from GeneRIF section in the NCBI RefSeq database. The dataset is downloaded from NCBI RefSeq Release 96, September 9, 2019. If you use this dataset please provide proper attribution to NCBI RefSeq and GlyGen. "
    ], 
    "_id": {
        "$oid": "5c8d4ff3f5c47b31a13ff81c"
    }, 
    "bco_id": "http://data.glygen.org/GLY_000133"
}