{
    "object_id": "https://biocomputeobject.org/GLY_000893/v-2.7.1",
    "etag": "a1f14502fd9553c1a25e8cc50aff4182",
    "spec_version": "1.3.0",
    "provenance_domain": {
        "name": "Pig Glycosylation Sites (GlyConnect) ",
        "version": "v-2.7.1",
        "license": "https://creativecommons.org/licenses/by/4.0/",
        "created": "",
        "modified": "2024-05-17T11:24:17.316Z",
        "contributors": [
            {
                "name": "Jeet Vora",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "jeetvora@gwu.edu"
            },
            {
                "name": "Robel Kahsay",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "rykahsay@gwu.edu"
            },
            {
                "name": "Frederique Lisacek",
                "affiliation": "GlyConnect",
                "contribution": [
                    "contributedBy"
                ],
                "email": "frederique.lisacek@sib.swiss"
            },
            {
                "name": "Julien Mariethoz",
                "affiliation": "GlyConnect",
                "contribution": [
                    "contributedBy"
                ],
                "email": "julien.mariethoz@sib.swiss"
            },
            {
                "name": "Karina Martinez",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "karinamartinez@gwu.edu"
            },
            {
                "name": "Jingyue Wu",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "jingyue.wu@email.gwu.edu"
            },
            {
                "name": "Kate Warner",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "k.warner1@gwu.edu"
            }
        ]
    },
    "usability_domain": [
        "The Pig Glycosylation Sites (GlyConnect) dataset contains Sus scrofa [taxid:9823] UniProtKB canonical accessions with information on reported glycans in GlyConnect database (https://glyconnect.expasy.org/; PMID: 30574787). A json file was retrieved through GlyConnect API (https://glyconnect.expasy.org/api/glycosylations?taxonomy=Sus%20scrofa)"
    ],
    "extension_domain": [
        {
            "extension_schema": "https://www.w3id.org/biocompute/extension_domain/1.2.0/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": "Sus scrofa",
                        "category_name": "species"
                    },
                    {
                        "category_value": "Proteoform",
                        "category_name": "molecule"
                    },
                    {
                        "category_value": "Glycosylation Sites",
                        "category_name": "tag"
                    },
                    {
                        "category_value": "csv",
                        "category_name": "file_type"
                    },
                    {
                        "category_value": "Reviewed",
                        "category_name": "status"
                    },
                    {
                        "category_value": "internal",
                        "category_name": "scope"
                    }
                ]
            }
        }
    ],
    "description_domain": {
        "keywords": [
            "glycosylation",
            "reported glycosylation"
        ],
        "pipeline_steps": [
            {
                "name": "make-proteoform-dataset.py",
                "version": "",
                "step_number": 1,
                "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/GLY_000001) and/or 2) if the given glycosylation position and amino acid does not match to the amino acid on the .fasta sequence provided in the latest UniProtKB release 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 (pig_proteoform_glycosylation_sites_glyconnect.1.log).",
                "prerequisite": [],
                "input_list": [],
                "output_list": []
            }
        ]
    },
    "execution_domain": {
        "script": [
            {
                "uri": {
                    "uri": "https://github.com/glygener/glygen-backend-integration/blob/master/dataset-maker/make-proteoform-dataset.py",
                    "filename": "make-proteoform-dataset.py"
                }
            }
        ],
        "script_driver": "manual",
        "software_prerequisites": [
            {
                "version": "2.7.5",
                "name": "Python",
                "uri": {
                    "uri": "https://www.python.org/download/releases/2.7.5/"
                }
            }
        ],
        "external_data_endpoints": [],
        "environment_variables": {}
    },
    "parametric_domain": [
        {
            "param": "",
            "value": "",
            "step": ""
        }
    ],
    "io_domain": {
        "output_subdomain": [
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.7.1/reviewed/pig_proteoform_glycosylation_sites_glyconnect.csv",
                    "filename": "pig_proteoform_glycosylation_sites_glyconnect.csv"
                }
            },
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.7.1/reviewed/pig_proteoform_glycosylation_sites_glyconnect.stat.csv",
                    "filename": "pig_proteoform_glycosylation_sites_glyconnect.stat.csv"
                }
            }
        ],
        "input_subdomain": [
            {
                "uri": {
                    "filename": "n_sequon_info.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.7.1/misc/n_sequon_info.csv"
                }
            },
            {
                "uri": {
                    "filename": "sample_mapping.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.7.1/misc/sample_mapping.csv"
                }
            },
            {
                "uri": {
                    "filename": "cellosaurus.txt",
                    "uri": "https:/data.glygen.org/ln2downloads/cellosaurus/2024_09_05/cellosaurus.txt"
                }
            },
            {
                "uri": {
                    "filename": "glyconnect_pig.json",
                    "uri": "https:/data.glygen.org/ln2downloads/glyconnect/2024_09_05/glyconnect_pig.json"
                }
            }
        ]
    },
    "error_domain": {
        "empirical_error": {},
        "algorithmic_error": {}
    }
}