{
    "io_domain": {
        "input_subdomain": [
            {
                "uri": {
                    "access_time": "", 
                    "sha1_chksum": "", 
                    "uri": "https://data.glygen.org//ln2downloads/glyconnect/09_09_2020/glyconnect_mouse_2020_06_08_10_41_58.json", 
                    "filename": "glyconnect_mouse_2020_06_08_10_41_58.json"
                }
            }, 
            {
                "uri": {
                    "access_time": "", 
                    "sha1_chksum": "", 
                    "uri": "https://data.glygen.org//ln2releases/data/v-1.7.13/reviewed/mouse_protein_canonicalsequences.fasta", 
                    "filename": "mouse_protein_canonicalsequences.fasta"
                }
            }, 
            {
                "uri": {
                    "access_time": "", 
                    "sha1_chksum": "", 
                    "uri": "https://data.glygen.org//ln2releases/data/v-1.7.13/reviewed/mouse_protein_masterlist.csv", 
                    "filename": "mouse_protein_masterlist.csv"
                }
            }
        ], 
        "output_subdomain": [
            {
                "mediatype": "csv", 
                "uri": {
                    "access_time": "2020-02-06T13:47:43-0500", 
                    "sha1_chksum": "", 
                    "uri": "http://data.glygen.org/ln2wwwdata/reviewed/mouse_proteoform_glycosylation_sites_glyconnect.csv", 
                    "filename": "mouse_proteoform_glycosylation_sites_glyconnect.csv"
                }
            }, 
            {
                "mediatype": "", 
                "uri": {
                    "access_time": "", 
                    "sha1_chksum": "", 
                    "uri": "http://data.glygen.org/ln2wwwdata/mouse_proteoform_glycosylation_sites_glyconnect.1.log", 
                    "filename": "mouse_proteoform_glycosylation_sites_glyconnect.1.log"
                }
            }, 
            {
                "mediatype": "csv", 
                "uri": {
                    "access_time": "", 
                    "sha1_chksum": "", 
                    "uri": "http://data.glygen.org/ln2wwwdata/reviewed/mouse_proteoform_glycosylation_sites_glyconnect.stat.csv", 
                    "filename": "mouse_proteoform_glycosylation_sites_glyconnect.stat.csv"
                }
            }
        ]
    }, 
    "bco_spec_version": "1.3.0", 
    "checksum": "aa993ad62e8d89164a5794b4471fd69b", 
    "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": "Proteoform", 
                "category_name": "molecule"
            }, 
            {
                "category_value": "Glycosylation Sites", 
                "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 Glycosylation Sites (GlyConnect) ", 
        "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": "GlyConnect", 
                "contribution": [
                    "authoredBy"
                ], 
                "name": "Frederique Lisacek", 
                "email": "frederique.lisacek@sib.swiss"
            }, 
            {
                "orcid": "", 
                "affiliation": "GlyConnect", 
                "contribution": [
                    "authoredBy"
                ], 
                "name": "Julien Mariethoz", 
                "email": "julien.mariethoz@sib.swiss"
            }
        ], 
        "created": "2020-02-10T12:16:18.548585", 
        "modified": "2020-08-07T15:07:41.538669", 
        "version": "1.7.13"
    }, 
    "description_domain": {
        "keywords": [
            "glycosylation", 
            "reported glycosylation"
        ], 
        "pipeline_steps": [
            {
                "name": "make-proteoform-dataset.py", 
                "step_number": 1, 
                "version": "", 
                "input_list": [
                    {
                        "access_time": "2020-08-07T15:00:35.734998", 
                        "uri": "https://data.glygen.org/ln2wwwdata/source/glyconnect_mouse_2020_06_08_10_41_58.json", 
                        "filename": "glyconnect_mouse_2020_06_08_10_41_58.json"
                    }, 
                    {
                        "access_time": "2020-08-07T15:00:35.734998", 
                        "uri": "https://data.glygen.org/ln2wwwdata/reviewed/mouse_protein_masterlist.csv", 
                        "filename": "mouse_protein_masterlist.csv"
                    }, 
                    {
                        "access_time": "2020-08-07T15:00:35.734998", 
                        "sha1_chksum": "", 
                        "uri": "https://data.glygen.org/ln2wwwdata/reviewed/mouse_protein_canonicalsequences.fasta", 
                        "filename": "mouse_protein_canonicalsequences.fasta"
                    }
                ], 
                "output_list": [
                    {
                        "access_time": "2020-08-07T15:00:35.734998", 
                        "uri": "https://data.glygen.org/ln2wwwdata/reviewed/mouse_proteoform_glycosylation_sites_glyconnect.csv", 
                        "filename": "mouse_proteoform_glycosylation_sites_glyconnect.csv"
                    }
                ], 
                "prerequisite": [], 
                "description": "The script loads the glycosylation information from the input JSON file and extracts the data. The QC checks filter out: 1) any rows where the UniProtKB accession is not included in the GlyGen protein list (https://data.glygen.org/GLYDS000001) and/or 2) if the given glycosylation position and amino acid does not match to the amino acid on the .fasta sequence for that protein (However it does pass if the amino acid +glyco-site is empty (as these records correspond to the \"not reported\" glyco-sites and/or 3) If the glycosylation type does not associate to the glycostlation based on amino acid (i.e if a N-linked is associated with Ser/Thr residue, etc). 4) If the glyco-site is not reported and the glycosylation type is ambigious (c-Linked/o-linked) and/or 5) if the saccharide (GlyTouCan) is not present in the GlyGen glycan masterlist. All the flagged entries are added in the log file (mouse_proteoform_glycosylation_sites_glyconnect.1.log)"
            }
        ]
    }, 
    "execution_domain": {
        "external_data_endpoints": [], 
        "environment_variables": {}, 
        "script_driver": "manual", 
        "software_prerequisites": [
            {
                "version": "2.7.5", 
                "name": "Python", 
                "uri": {
                    "access_time": "2020-08-07T15:00:35.734998", 
                    "sha1_chksum": "", 
                    "uri": "https://www.python.org/download/releases/2.7.5/"
                }
            }
        ], 
        "script": [
            {
                "uri": {
                    "access_time": "2020-08-07T15:00:35.734998", 
                    "sha1_chksum": "", 
                    "uri": "https://github.com/GW-HIVE/glygen-backend-integration/blob/master/pipeline/integrator/make-proteoform-dataset.py", 
                    "filename": "make-proteoform-dataset.py"
                }
            }
        ]
    }, 
    "error_domain": {
        "empirical_error": {}, 
        "algorithmic_error": {}, 
        "emperical_error": {}
    }, 
    "parametric_domain": [
        {
            "step": "", 
            "value": "", 
            "param": ""
        }
    ], 
    "usability_domain": [
        "The Mouse Glycosylation Sites (GlyConnect) dataset contains mouse [taxid:10090] UniProtKB canonical accessions with information on reported glycosylation sites in GlyConnect database (https://glyconnect.expasy.org/; PMID: 30574787).A json file was derived through GlyConnect API (https://glyconnect.expasy.org/api/docs/index.html?url=/api/assets/swagger.json#/Glycosylation/listGlycosylationsAsJson) (Query: taxonomy:Mus musculus). If you use this dataset please provide proper attribution to GlyConnect and GlyGen."
    ], 
    "_id": {
        "$oid": "5e418fe2f5c47b111cca84ff"
    }, 
    "bco_id": "http://data.glygen.org/GLY_000330"
}