{
    "io_domain": {
        "input_subdomain": [
            {
                "uri": {
                    "access_time": "", 
                    "sha1_chksum": "", 
                    "uri": "https://data.glygen.org//ln2releases/data/v-1.5.36/reviewed/sarscov2_protein_masterlist.csv", 
                    "filename": "sarscov2_protein_masterlist.csv"
                }
            }, 
            {
                "uri": {
                    "access_time": "", 
                    "sha1_chksum": "", 
                    "uri": "https://data.glygen.org//ln2downloads/ebi/2020_05/uniprot-proteome-sars-cov-2.nt", 
                    "filename": "uniprot-proteome-sars-cov-2.nt"
                }
            }
        ], 
        "output_subdomain": [
            {
                "mediatype": "csv", 
                "uri": {
                    "access_time": "", 
                    "uri": "https://data.glygen.org/ln2wwwdata/reviewed/sarscov2_protein_signalp_annotation.csv", 
                    "filename": "sarscov2_protein_signalp_annotation.csv"
                }
            }, 
            {
                "mediatype": "fasta", 
                "uri": {
                    "access_time": "", 
                    "uri": "https://data.glygen.org/ln2wwwdata/reviewed/sarscov2_protein_signalp_peptidesequences.fasta", 
                    "filename": "sarscov2_protein_signalp_peptidesequences.fasta"
                }
            }, 
            {
                "mediatype": "fasta", 
                "uri": {
                    "access_time": "", 
                    "uri": "https://data.glygen.org/ln2wwwdata/reviewed/sarscov2_protein_signalp_cleavedsequences.fasta", 
                    "filename": "sarscov2_protein_signalp_cleavedsequences.fasta"
                }
            }, 
            {
                "mediatype": "fasta", 
                "uri": {
                    "access_time": "", 
                    "sha1_chksum": "", 
                    "uri": "https://data.glygen.org/ln2wwwdata/reviewed/sarscov2_protein_signalp_fullsequences.fasta", 
                    "filename": "sarscov2_protein_signalp_fullsequences.fasta"
                }
            }, 
            {
                "mediatype": "csv", 
                "uri": {
                    "access_time": "", 
                    "sha1_chksum": "", 
                    "uri": "https://data.glygen.org/ln2wwwdata/reviewed/sarscov2_protein_signalp_annotation.stat.csv", 
                    "filename": "sarscov2_protein_signalp_annotation.stat.csv"
                }
            }
        ]
    }, 
    "bco_spec_version": "1.3.0", 
    "checksum": "56a2fb1145448a792dcc447ac5d507e3", 
    "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": "SARS coronavirus (SARS-CoV-2 or 2019-nCoV)", 
                "category_name": "species"
            }, 
            {
                "category_value": "Protein", 
                "category_name": "molecule"
            }, 
            {
                "category_value": "Signal Peptide", 
                "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": "SARS-CoV2 Protein Signal Peptide", 
        "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": "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"
            }
        ], 
        "created": "2020-04-03T11:48:23.646763", 
        "modified": "2020-04-10T12:28:50.758582", 
        "version": "1.5.36"
    }, 
    "description_domain": {
        "keywords": [
            "signal peptide", 
            "ptm processing", 
            "UniProtKB"
        ], 
        "pipeline_steps": [
            {
                "name": "make-protein-dataset.py", 
                "step_number": 1, 
                "version": "", 
                "input_list": [
                    {
                        "access_time": "", 
                        "uri": "https://data.glygen.org/ln2wwwdata/reviewed/sarscov2_protein_masterlist.csv", 
                        "filename": "sarscov2_protein_masterlist.csv"
                    }, 
                    {
                        "access_time": "", 
                        "uri": "https://data.glygen.org/ln2wwwdata/source/uniprot-proteome-sars-cov-2.nt", 
                        "filename": "uniprot-proteome-sars-cov-2.nt"
                    }
                ], 
                "output_list": [
                    {
                        "access_time": "", 
                        "uri": "https://data.glygen.org/ln2wwwdata/reviewed/sarscov2_protein_signalp_annotation.csv", 
                        "filename": "sarscov2_protein_signalp_annotation.csv"
                    }, 
                    {
                        "access_time": "", 
                        "uri": "https://data.glygen.org/ln2wwwdata/reviewed/sarscov2_protein_signalp_peptidesequences.fasta", 
                        "filename": "sarscov2_protein_signalp_peptidesequences.fasta"
                    }, 
                    {
                        "access_time": "", 
                        "uri": "https://data.glygen.org/ln2wwwdata/reviewed/sarscov2_protein_signalp_cleavedsequences.fasta", 
                        "filename": "sarscov2_protein_signalp_cleavedsequences.fasta"
                    }, 
                    {
                        "access_time": "", 
                        "sha1_chksum": "", 
                        "uri": "https://data.glygen.org/ln2wwwdata/reviewed/sarscov2_protein_signalp_fullsequences.fasta", 
                        "filename": "sarscov2_protein_signalp_fullsequences.fasta"
                    }
                ], 
                "prerequisite": [], 
                "description": "The script extracts the signal peptide information from the triplestore and maps 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": "", 
                    "sha1_chksum": "", 
                    "uri": "https://www.python.org/download/releases/2.7.5/"
                }
            }
        ], 
        "script": [
            {
                "uri": {
                    "access_time": "", 
                    "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 SARS coronavirus (SARS-CoV-2 or 2019-nCov) Protein Signal Peptide dataset contains sarscov2 [taxid:2697049] signal peptide information for the UniProtKB canonical accessions. The dataset contains protein information, signal peptide sequence, signal peptide length, signal peptide  start and end position, cleaved protein sequence, eco ids and pmids. The dataset contains related files in the output subdomain that includes log file containing accessions that do not have signal peptide information and processed protein sequence\u2019s files in fasta format. The dataset is derived from 2020-01 UniProtKB release. If you use this dataset please provide proper attribution to EMBL-EBI-UniProtKB and GlyGen."
    ], 
    "_id": {
        "$oid": "5e875ac7f5c47b6da3059efe"
    }, 
    "bco_id": "http://data.glygen.org/GLY_000443"
}