{
    "io_domain": {
        "input_subdomain": [
            {
                "uri": {
                    "access_time": "", 
                    "sha1_chksum": "", 
                    "uri": "https://data.glygen.org//ln2downloads/ebi/2020_05/uniprot-proteome-mus-musculus.nt", 
                    "filename": "uniprot-proteome-mus-musculus.nt"
                }
            }
        ], 
        "output_subdomain": [
            {
                "mediatype": "csv", 
                "uri": {
                    "access_time": "2019-10-26T17:45:17-0500", 
                    "sha1_chksum": "", 
                    "uri": "http://data.glygen.org/ln2wwwdata/reviewed/mouse_protein_transcriptlocus.csv", 
                    "filename": "mouse_protein_transcriptlocus.csv"
                }
            }, 
            {
                "mediatype": "csv", 
                "uri": {
                    "access_time": "2019-10-26T17:45:17-0500", 
                    "sha1_chksum": "", 
                    "uri": "http://data.glygen.org/ln2wwwdata/reviewed/mouse_protein_transcriptlocus.stat.csv", 
                    "filename": "mouse_protein_transcriptlocus.stat.csv"
                }
            }
        ]
    }, 
    "bco_spec_version": "1.3.0", 
    "checksum": "346d83a0b5d6e2f8a978339f3e5bcac9", 
    "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": "Transcript Coordinates", 
                "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 Transcript Locus (Ensembl Transcript coordinates)", 
        "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"
            }
        ], 
        "created": "2019-03-16T16:06:41.833239", 
        "modified": "2020-04-10T12:44:17.413281", 
        "version": "1.7.13"
    }, 
    "description_domain": {
        "keywords": [
            "transcript locus", 
            "ensembl transcript id", 
            "ensembl peptide id", 
            "chromosome", 
            "Ensembl"
        ], 
        "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/mouse_protein_masterlist.csv", 
                        "filename": "mouse_protein_masterlist.csv"
                    }, 
                    {
                        "access_time": "2019-10-26T17:43:17-0500", 
                        "uri": "https://data.glygen.org/ln2wwwdata/source/uniprot-proteome-mus-musculus.nt", 
                        "filename": "uniprot-proteome-mus-musculus.nt"
                    }
                ], 
                "output_list": [
                    {
                        "access_time": "2019-10-26T17:45:17-0500", 
                        "uri": "https://data.glygen.org/ln2wwwdata/reviewed/mouse_protein_transcriptlocus.csv", 
                        "filename": "mouse_protein_transcriptlocus.csv"
                    }
                ], 
                "prerequisite": [], 
                "description": "The script loads the UniProtKB accessions that are mapped to UniProtKB isoforms and ENSEMBL Transcript and Peptide identifiers from the triplestore containing triples from nt file and maps it to the UniProtKB canonical accession to finally extract the data as a csv dataset"
            }
        ]
    }, 
    "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": [
        "Th Mouse Transcript Locus (Ensembl Transcript coordinates)contains UniProtKB mouse [taxid:10090] canonical accessions mapped to their isoforms with further mapping to the corresponding ENSEMBL Transcript ID, Peptide ID, chromosome ID along with the transcripts's start and end position. The dataset is derived from 2019-09 UniProtKB release. If you use this dataset please provide proper attribution to EMBL-EBI-UniProtKB and GlyGen"
    ], 
    "_id": {
        "$oid": "5c8d5751f5c47b34a28a481b"
    }, 
    "bco_id": "http://data.glygen.org/GLY_000136"
}