{
    "object_id": "https://biocomputeobject.org/GLY_000623/v-2.12.1",
    "spec_version": "1.3.0",
    "etag": "edd652a836dce14de22f8b4b99fa4b65",
    "provenance_domain": {
        "name": "Human Glycogenes",
        "version": "v-2.12.1",
        "license": "https://creativecommons.org/licenses/by/4.0/",
        "created": "2022-03-29T10:11:35.122210",
        "modified": "2026-03-26T15:04:51.908Z",
        "contributors": [
            {
                "name": "Jeet Vora",
                "affiliation": "The George Washington University",
                "email": "jeetvora@gwu.edu",
                "contribution": [
                    "createdBy"
                ]
            },
            {
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "name": "Robel Kahsay",
                "email": "rykahsay@gwu.edu"
            },
            {
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "name": "Karina Martinez",
                "email": "karinamartinez@gwu.edu"
            },
            {
                "name": "Luke Johnson",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "luke.johnson@email.gwu.edu"
            },
            {
                "name": "Urnisha Bhuiyan",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "sbhuiyan28@email.gwu.edu"
            },
            {
                "name": "Cyrus Au-Yeung",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "cyruschauyeung@gwu.edu"
            },
            {
                "name": "Kate Warner",
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "email": "k.warner1@gwu.edu"
            },
            {
                "name": "Kelley Moremen",
                "affiliation": "University of Georgia",
                "email": "moremen@uga.edu",
                "contribution": [
                    "createdBy"
                ]
            }
        ]
    },
    "usability_domain": [
        "This Human glycogenes dataset is retrieved a publication from UGA [PMID:18411279]. It contains a list of UniProtKB accessions full and shortened human [taxonomy:9606] protein names mapped to NCBI gene and HGNC IDs. The GT, GH, CBM and CE groups are defined by CAZy families. Following those groups are Lectins (following Kurt Drickamer\u2019s groupings), then several other glycan-related genes like GL (Glycolipid-related), GR (GAG-related), etc. If you use this dataset, please provide proper attribution to RefSeq, EMBL-EBI-UniProtKB and GlyGen. Cite:PMID:26553804, PMID:33237286 and PMID:31616925. The dataset can be used for obtaining UniProtKB recommended names (full and short) and HGNC gene names for UniProtKB canonical proteins. If you use this dataset, please provide proper attribution to EMBL-EBI-UniProtKB and GlyGen. This dataset can be used for obtaining proteins grouped as they are defined by CAZy."
    ],
    "description_domain": {
        "pipeline_steps": [
            {
                "name": "make-protein-dataset.py",
                "step_number": 1,
                "input_list": [],
                "description": "The script maps the gene ids and gene names from the source file to UniProtKB canonical accession and then then adds the protein recommneded name and alternative gene names. The script extracts gene identifier and group fields as is from the source input file. The script filters entries that are not mapped in the log file.",
                "prerequisite": [],
                "output_list": []
            }
        ],
        "keywords": [
            "glycogenes"
        ]
    },
    "parametric_domain": [
        {
            "value": ""
        }
    ],
    "io_domain": {
        "output_subdomain": [
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.12.1/reviewed/human_protein_glycogenes.csv",
                    "filename": "human_protein_glycogenes.csv"
                }
            },
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.12.1/reviewed/human_protein_glycogenes.stat.csv",
                    "filename": "human_protein_glycogenes.stat.csv"
                }
            }
        ],
        "input_subdomain": [
            {
                "uri": {
                    "filename": "human_protein_xref_geneid.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.12.1/reviewed/human_protein_xref_geneid.csv"
                }
            },
            {
                "uri": {
                    "filename": "human_glycogenes.csv",
                    "uri": "https:/data.glygen.org/ln2downloads/glycogenes/2024_02_14/human_glycogenes.csv"
                }
            }
        ]
    },
    "execution_domain": {
        "script": [
            {
                "uri": {
                    "uri": "https://github.com/glygener/glygen-backend-integration/blob/master/dataset-maker/make-protein-dataset.py",
                    "filename": "make-protein-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": {}
    },
    "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": "Homo sapiens",
                        "category_name": "species"
                    },
                    {
                        "category_value": "Protein",
                        "category_name": "molecule"
                    },
                    {
                        "category_value": "Glycogenes",
                        "category_name": "tag"
                    },
                    {
                        "category_value": "csv",
                        "category_name": "file_type"
                    },
                    {
                        "category_value": "Reviewed",
                        "category_name": "status"
                    },
                    {
                        "category_value": "internal",
                        "category_name": "scope"
                    },
                    {
                        "category_name": "file_category",
                        "category_value": "others"
                    }
                ]
            }
        }
    ],
    "error_domain": {}
}