{
    "object_id": "https://biocomputeobject.org/GLY_000035/1.0.13", 
    "license": "https://creativecommons.org/licenses/by/4.0/", 
    "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": [], 
            "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": "Leyla Garcia", 
            "email": "ljgarcia@ebi.ac.uk"
        }, 
        {
            "orcid": "", 
            "affiliation": "EMBL-EBI-UniProt", 
            "contribution": [
                "contributedBy"
            ], 
            "name": "Preethi Vasudev", 
            "email": "pvasudev@ebi.ac.uk"
        }
    ], 
    "created": "2019-01-23T17:21:05.902655", 
    "spec_version": "1.3.0", 
    "modified": "2019-01-25T16:52:04.119358", 
    "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": "Reviewed", 
                        "category_name": "status"
                    }
                ]
            }
        }
    ], 
    "io_domain": {
        "input_subdomain": [
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2wwwdata/source/uniprot-proteome-9606-homo-sapiens-2018-11.nt", 
                    "filename": "uniprot-proteome-9606-homo-sapiens-2018-11.nt"
                }
            }, 
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2wwwdata/reviewed/human_protein_blacklisted_pmids.csv", 
                    "filename": "human_protein_blacklisted_pmids.csv"
                }
            }, 
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2wwwdata/reviewed/human_protein_idmapping.csv", 
                    "filename": "human_protein_idmapping.csv"
                }
            }
        ], 
        "output_subdomain": [
            {
                "mediatype": "csv", 
                "uri": {
                    "uri": "https://data.glygen.org/ln2wwwdata/reviewed/human_protein_citations.csv", 
                    "filename": "human_protein_citations.csv"
                }
            }, 
            {
                "mediatype": "csv", 
                "uri": {
                    "uri": "https://data.glygen.org/ln2wwwdata/reviewed/human_protein_citations.stat.csv", 
                    "filename": "human_protein_citations.stat.csv"
                }
            }
        ]
    }, 
    "provenance_domain": {
        "license": "https://creativecommons.org/licenses/by/4.0/", 
        "name": "Human Protein Citations (UniProtKB)", 
        "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": "EMBL-EBI-UniProt", 
                "contribution": [
                    "contributedBy"
                ], 
                "name": "Maria Martin", 
                "email": "martin@ebi.ac.uk"
            }, 
            {
                "affiliation": "EMBL-EBI-UniProt", 
                "contribution": [
                    "contributedBy"
                ], 
                "name": "Leyla Garcia", 
                "email": "ljgarcia@ebi.ac.uk"
            }, 
            {
                "affiliation": "EMBL-EBI-UniProt", 
                "contribution": [
                    "contributedBy"
                ], 
                "name": "Preethi Vasudev", 
                "email": "pvasudev@ebi.ac.uk"
            }
        ], 
        "created": "2019-01-23T17:21:05.902655", 
        "modified": "2019-04-25T19:01:56.591862", 
        "version": "1.0.13"
    }, 
    "description_domain": {
        "keywords": [
            "publications", 
            "PMID", 
            "journal", 
            "citations"
        ], 
        "pipeline_steps": [
            {
                "name": "extract-from-endpoint.py", 
                "step_number": 1, 
                "version": "1.0", 
                "input_list": [
                    {
                        "uri": "https://data.glygen.org/ln2wwwdata/source/uniprot-proteome-9606-homo-sapiens-2018-11.nt", 
                        "filename": "uniprot-proteome-9606-homo-sapiens-2018-11.nt"
                    }, 
                    {
                        "uri": "https://data.glygen.org/ln2wwwdata/reviewed/human_protein_blacklisted_pmids.csv", 
                        "filename": "human_protein_blacklisted_pmids.csv"
                    }, 
                    {
                        "uri": "https://data.glygen.org/ln2wwwdata/reviewed/human_protein_idmapping.csv", 
                        "filename": "human_protein_idmapping.csv"
                    }
                ], 
                "output_list": [
                    {
                        "uri": "https://data.glygen.org/ln2wwwdata/reviewed/human_protein_citations.csv", 
                        "filename": "human_protein_citations.csv"
                    }
                ], 
                "prerequisite": [], 
                "description": "The script loads the UniProtKB publication data from the triplestore containing triples from nt file to retrieve the publication information associated with the UniProtKB accession and maps it to the UniProtKB canonical accessions. The script also filter out publications of large scale studies, and high frequency publications pertaining to DNA sequencing, cDNA, chromosomes"
            }
        ]
    }, 
    "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/glygener/glygen-backend-integration/blob/master/pipeline/extractor/extract-from-endpoint.py", 
                    "filename": "extract-from-endpoint.py"
                }
            }
        ]
    }, 
    "version": "1.0.5", 
    "error_domain": {
        "empirical_error": {}, 
        "algorithmic_error": {}, 
        "emperical_error": {}
    }, 
    "parametric_domain": [
        {
            "step": "", 
            "value": "", 
            "param": ""
        }
    ], 
    "usability_domain": [
        "The Human Protein Citations dataset contains publication information for human [taxid:9606] UniProtKB accessions that includes PMID, journal name and title. Publications of large scale studies, and high frequency publications pertaining to genome sequencing, cDNA, chromosomes have been filtered out as per https://data.glygen.org/ln2wwwdata/reviewed/human_protein_blacklisted_pmids.csv", 
        "The dataset is derived from 2018-11 UniProt release. If you use this dataset please provide proper attribution to EMBL-EBI-UniProt and GlyGen"
    ], 
    "etag": "6050e29a09e9b669ec18385d5b994da9", 
    "_id": {
        "$oid": "5c48e8d1f5c47b65ae4b5d0e"
    }, 
    "name": "Human Protein Citations"
}