{
    "object_id": "https://biocomputeobject.org/GLY_000043/v-1.0.13",
    "license": "https://creativecommons.org/licenses/by/4.0/",
    "contributors": [
        {
            "orcid": "",
            "affiliation": "George Washington University",
            "contribution": [
                "createdBy"
            ],
            "name": "Rahi Navelkar",
            "email": "rsn13@gwu.edu"
        },
        {
            "orcid": "",
            "affiliation": "RCSB PDB",
            "contribution": [
                "authoredBy"
            ],
            "name": "Jasmine Young ",
            "email": "jasmine@rcsb.rutgers.edu"
        },
        {
            "orcid": "",
            "affiliation": "RCSB PDB",
            "contribution": [
                "authoredBy"
            ],
            "name": "Chenghua Shao ",
            "email": "chenghua@rcsb.rutgers.edu"
        },
        {
            "orcid": "",
            "affiliation": "George Washington University",
            "contribution": [
                "contributedBy"
            ],
            "name": "Nagarajan Pattabiraman",
            "email": "pattabir@email.gwu.edu"
        },
        {
            "orcid": "",
            "affiliation": "George Washington University",
            "contribution": [
                "contributedBy"
            ],
            "name": "Reza Mousavi ",
            "email": "rezamousavi@gwmail.gwu.edu"
        }
    ],
    "created": "2019-01-23T17:21:05.910839",
    "spec_version": "1.3.0",
    "modified": "2019-02-20T11:26:26.174289",
    "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": "source/N-linked-ASN.txt",
                    "filename": "N-linked-ASN.txt"
                }
            },
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2data/releases/data/v-1.0.13/source/O-linked-SER.txt",
                    "filename": "O-linked-SER.txt"
                }
            },
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2data/releases/data/v-1.0.13/source/O-linked-THR.txt",
                    "filename": "https://data.glygen.org/ln2data/releases/data/v-1.0.13/source/O-linked-THR.txt"
                }
            },
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2data/releases/data/v-1.0.13/reviewed/mouse_protein_all.fasta",
                    "filename": "mouse_protein_all.fasta"
                }
            }
        ],
        "output_subdomain": [
            {
                "mediatype": "",
                "uri": {
                    "uri": "https://data.glygen.org/ln2data/releases/data/v-1.0.13/reviewed/mouse_proteoform_glycosylation_sites_pdb.csv",
                    "filename": "mouse_proteoform_glycosylation_sites_pdb.csv"
                }
            },
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https://data.glygen.org/ln2data/releases/data/v-1.0.13/reviewed/mouse_proteoform_glycosylation_sites_pdb.stat.csv",
                    "filename": "mouse_proteoform_glycosylation_sites_pdb.stat.csv"
                }
            }
        ]
    },
    "provenance_domain": {
        "license": "https://creativecommons.org/licenses/by/4.0/",
        "name": "Mouse Glycosylation Sites (RCSB PDB)",
        "contributors": [
            {
                "affiliation": "George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "name": "Rahi Navelkar",
                "email": "rsn13@gwu.edu"
            },
            {
                "affiliation": "George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "name": "Jeet Vora",
                "email": "jeetvora@gwu.edu"
            },
            {
                "affiliation": "RCSB PDB",
                "contribution": [
                    "authoredBy"
                ],
                "name": "Jasmine Young ",
                "email": "jasmine@rcsb.rutgers.edu"
            },
            {
                "affiliation": "RCSB PDB",
                "contribution": [
                    "authoredBy"
                ],
                "name": "Chenghua Shao ",
                "email": "chenghua@rcsb.rutgers.edu"
            },
            {
                "affiliation": "George Washington University",
                "contribution": [
                    "contributedBy"
                ],
                "name": "Nagarajan Pattabiraman",
                "email": "pattabir@email.gwu.edu"
            },
            {
                "affiliation": "George Washington University",
                "contribution": [
                    "contributedBy"
                ],
                "name": "Reza Mousavi ",
                "email": "rezamousavi@gwmail.gwu.edu"
            }
        ],
        "created": "2019-01-23T17:21:05.910839",
        "modified": "2019-04-11T13:56:50.790064",
        "version": "v-1.0.13"
    },
    "description_domain": {
        "keywords": [
            "proteins",
            "canonical",
            "glycosylation"
        ],
        "pipeline_steps": [
            {
                "name": "python",
                "step_number": 1,
                "version": "",
                "input_list": [
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-1.0.13/source/N-linked-ASN.txt",
                        "filename": "N-linked-ASN.txt"
                    },
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-1.0.13/source/O-linked-SER.txt",
                        "filename": "O-linked-SER.txt"
                    },
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-1.0.13/source/O-linked-THR.txt",
                        "filename": "O-linked-THR.txt"
                    }
                ],
                "output_list": [],
                "prerequisite": [],
                "description": "The list of PDB Id's was retrieved from the source and their PDB header files were downloaded on september 27, 2018 and the header files were parsed to extract information on glycosylation positions for PDB Id's and UniProt accessions using headers 'DBREF', 'LINK' and 'MODRES' using a script [make-proteoform-glycosylation_sites_pdb_step1-csv.py]."
            },
            {
                "name": "python",
                "step_number": 1,
                "version": "",
                "input_list": [],
                "output_list": [
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-1.0.13/logs/proteoform_glycosylation_sites_pdb.log",
                        "filename": "proteoform_glycosylation_sites_pdb.log"
                    }
                ],
                "prerequisite": [],
                "description": "PDB Ids where information in the header files was either not available or was incorrect were eliminated. [The eliminated PDB Id's can be accessed through log file- proteoform_glycosylation_sites_pdb.log in output subdomain]."
            },
            {
                "name": "python",
                "step_number": 2,
                "version": "",
                "input_list": [
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-1.0.13/reviewed/mouse_protein_all.fasta",
                        "filename": "mouse_protein_all.fasta"
                    }
                ],
                "output_list": [
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-1.0.13/logs/mouse_proteoform_glycosylation_sites_pdb.log",
                        "filename": "mouse_proteoform_glycosylation_sites_pdb.log"
                    },
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-1.0.13/reviewed/mouse_proteoform_glycosylation_sites_pdb.csv",
                        "filename": "mouse_proteoform_glycosylation_sites_pdb.csv"
                    }
                ],
                "prerequisite": [],
                "description": "Records which fall under one or more following conditions were flagged and eliminated. [The eliminated records can be accessed through log file- mouse_proteoform_glycosylation_sites_pdb.log in output subdomain]:  a. A mismatch between 'amino_acid_pdb' and 'amino_acid_uniprotkb' is observed. b. A mismatch between 'amino_acid_uniprotkb' and amino acid retrieved from fasta sequences (UniProtKB NOV-17 version) is observed. "
            }
        ]
    },
    "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/blob/master/integration/",
                    "filename": "make-proteoform-glycosylation_sites_pdb_step1-csv.py"
                }
            },
            {
                "uri": {
                    "uri": "https://github.com/glygener/glygen-backend/blob/master/integration/",
                    "filename": "ac2canonical.py"
                }
            },
            {
                "uri": {
                    "uri": "https://github.com/glygener/glygen-backend/blob/master/integration/",
                    "filename": "make-proteoform-glycosylation_sites_pdb-csv.py"
                }
            }
        ]
    },
    "version": "1.0.5",
    "error_domain": {
        "empirical_error": {},
        "algorithmic_error": {},
        "emperical_error": {}
    },
    "parametric_domain": [
        {
            "step": "",
            "value": "",
            "param": ""
        }
    ],
    "usability_domain": [
        "List of mouse [taxid:10090] proteins with information on glycosylation sites from UniCarbKB database. [https://academic.oup.com/nar/article/42/D1/D215/1052197 ,https://doi.org/10.1093/nar/gkt1128]"
    ],
    "etag": "32901127e1c86c234aa3614085a2b38a",
    "_id": {
        "$oid": "5c48e8d1f5c47b65ae4b5d16"
    },
    "name": "Mouse Glycosylation Sites (RCSB PDB)"
}