{
    "io_domain": {
        "input_subdomain": [
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2wwwdata/source/N-linked-ARG", 
                    "filename": "N-linked-ARG"
                }
            }, 
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2wwwdata/source/pdb_chain_uniprot_2018_11_8_11_44_50.csv", 
                    "filename": "pdb_chain_uniprot_2018_11_8_11_44_50.csv"
                }
            }, 
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2wwwdata/source/O-linked-ASP", 
                    "filename": "O-linked-ASP"
                }
            }, 
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2wwwdata/source/O-linked-GLU", 
                    "filename": "O-linked-GLU"
                }
            }, 
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2wwwdata/source/O-linked-SER", 
                    "filename": "O-linked-SER"
                }
            }, 
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2wwwdata/source/O-linked-THR", 
                    "filename": "O-linked-THR"
                }
            }, 
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2wwwdata/source/O-linked-TYR", 
                    "filename": "O-linked-TYR"
                }
            }, 
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2wwwdata/reviewed/rat_protein_masterlist.csv", 
                    "filename": "rat_protein_masterlist.csv"
                }
            }
        ], 
        "output_subdomain": [
            {
                "mediatype": "csv", 
                "uri": {
                    "uri": "http://data.glygen.org/ln2wwwdata/reviewed/rat_proteoform_glycosylation_sites_pdb.csv", 
                    "filename": "rat_proteoform_glycosylation_sites_pdb.csv"
                }
            }, 
            {
                "mediatype": "csv", 
                "uri": {
                    "uri": "http://data.glygen.org/ln2wwwdata/reviewed/rat_proteoform_glycosylation_sites_pdb.stat.csv", 
                    "filename": "rat_proteoform_glycosylation_sites_pdb.stat.csv"
                }
            }
        ]
    }, 
    "spec_version": "1.3.0", 
    "object_id": "https://biocomputeobject.org/GLY_000226/1.5.11", 
    "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": "Rattus Norvegicus", 
                        "category_name": "species"
                    }, 
                    {
                        "category_value": "Protein", 
                        "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": "Rat Glycosylation Sites (RCSB PDB)", 
        "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": "The George Washington University", 
                "contribution": [
                    "createdBy"
                ], 
                "name": "Xavier Holmes", 
                "email": "xavierh@email.gwu.edu"
            }, 
            {
                "affiliation": "RCSB PDB", 
                "contribution": [
                    "contributedBy"
                ], 
                "name": "Jasmine Young", 
                "email": "jasmine@rcsb.rutgers.edu"
            }, 
            {
                "affiliation": "RCSB PDB", 
                "contribution": [
                    "contributedBy"
                ], 
                "name": "Chenghua Shao", 
                "email": "chenghua@rcsb.rutgers.edu"
            }
        ], 
        "created": "2019-09-06T14:06:38.954510", 
        "modified": "2019-11-26T16:36:08.184339", 
        "version": "1.5.11"
    }, 
    "description_domain": {
        "keywords": [
            "glycosylation", 
            "reported glycosylation", 
            "predicted glycosylation", 
            "UniProtKB", 
            "PubMed", 
            "Evidence and conclusion Ontology"
        ], 
        "pipeline_steps": [
            {
                "name": "make-proteoform-dataset.py", 
                "step_number": 1, 
                "version": "", 
                "input_list": [
                    {
                        "uri": "https://data.glygen.org/ln2wwwdata/source/N-linked-ARG", 
                        "filename": "N-linked-ARG"
                    }, 
                    {
                        "uri": "https://data.glygen.org/ln2wwwdata/source/N-linked-ASN", 
                        "filename": "N-linked-ASN"
                    }, 
                    {
                        "uri": "https://data.glygen.org/ln2wwwdata/source/O-linked-ASP", 
                        "filename": "O-linked-ASP"
                    }, 
                    {
                        "uri": "https://data.glygen.org/ln2wwwdata/source/O-linked-GLU", 
                        "filename": "O-linked-GLU"
                    }, 
                    {
                        "uri": "https://data.glygen.org/ln2wwwdata/source/O-linked-SER", 
                        "filename": "O-linked-SER"
                    }, 
                    {
                        "uri": "https://data.glygen.org/ln2wwwdata/source/O-linked-THR", 
                        "filename": "O-linked-THR"
                    }, 
                    {
                        "uri": "https://data.glygen.org/ln2wwwdata/source/O-linked-TYR", 
                        "filename": "O-linked-TYR"
                    }, 
                    {
                        "uri": "https://data.glygen.org/ln2wwwdata/reviewed/rat_protein_masterlist.csv", 
                        "filename": "rat_protein_masterlist.csv"
                    }
                ], 
                "output_list": [
                    {
                        "uri": "https://data.glygen.org/ln2wwwdata/reviewed/rat_proteoform_glycosylation_sites_pdb.csv", 
                        "filename": "rat_proteoform_glycosylation_sites_pdb.csv"
                    }
                ], 
                "prerequisite": [], 
                "description": "The script extracts the PDB IDs from the N-linked and O-linked files and extract the PDB files containing the glycosylation information from RCSB PDB using their API. The script using also checks the position of UniProtKB aa position in the file with the current sequences and finally maps the UniProtKB accession to the UniProtKB canonical accessions for the given organism."
            }
        ]
    }, 
    "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/GW-HIVE/glygen-backend-integration/blob/master/pipeline/integrator/make-proteoform-dataset.py", 
                    "filename": "make-proteoform-dataset.py"
                }
            }
        ]
    }, 
    "usability_domain": [
        "The Rat Glycosylation Sites (RCSB PDB) dataset contains rat [taxid:10116] UniProtKB canonical accessions with information on reported and predicted glycosylation sites in RCSB PDB database. If you use this dataset please provide proper attribution to RCSB PDB and GlyGen"
    ], 
    "etag": "43b7cc2fe8ed801ba73404a6a97b7a5c", 
    "_id": {
        "$oid": "5d72a02ef5c47b0f43423353"
    }
}