{
    "io_domain": {
        "input_subdomain": [
            {
                "uri": {
                    "filename": "cellosaurus.txt",
                    "uri": "https:/data.glygen.org/ln2downloads/cellosaurus/2020-09-15/cellosaurus.txt"
                }
            },
            {
                "uri": {
                    "filename": "known_legacy_dbogap_dbogap2_glygen.csv",
                    "uri": "https:/data.glygen.org/ln2downloads/unicarbkb/01_11_2022/known_legacy_dbogap_dbogap2_glygen.csv"
                }
            },
            {
                "uri": {
                    "filename": "known_34106099_glygen.csv",
                    "uri": "https:/data.glygen.org/ln2downloads/unicarbkb/01_11_2022/known_34106099_glygen.csv"
                }
            },
            {
                "uri": {
                    "filename": "known_unicarbkb_glycomics_study.csv",
                    "uri": "https:/data.glygen.org/ln2downloads/unicarbkb/01_11_2022/known_unicarbkb_glycomics_study.csv"
                }
            },
            {
                "uri": {
                    "filename": "fuzzy_legacy_cph_start_end_positions.csv",
                    "uri": "https:/data.glygen.org/ln2downloads/unicarbkb/01_11_2022/fuzzy_legacy_cph_start_end_positions.csv"
                }
            },
            {
                "uri": {
                    "filename": "known_legacy_cph_single_positions.csv",
                    "uri": "https:/data.glygen.org/ln2downloads/unicarbkb/01_11_2022/known_legacy_cph_single_positions.csv"
                }
            },
            {
                "uri": {
                    "filename": "unknown_legacy_human_proteoform_glycosylation_sites_unicarbkb_ogalnac_data.csv",
                    "uri": "https:/data.glygen.org/ln2downloads/unicarbkb/01_11_2022/unknown_legacy_human_proteoform_glycosylation_sites_unicarbkb_ogalnac_data.csv"
                }
            },
            {
                "uri": {
                    "filename": "known_legacy_human_mouse_rat_glygen.csv",
                    "uri": "https:/data.glygen.org/ln2downloads/unicarbkb/01_11_2022/known_legacy_human_mouse_rat_glygen.csv"
                }
            },
            {
                "uri": {
                    "filename": "known_34004174_glygen.csv",
                    "uri": "https:/data.glygen.org/ln2downloads/unicarbkb/01_11_2022/known_34004174_glygen.csv"
                }
            },
            {
                "uri": {
                    "filename": "known_33806155_glygen.csv",
                    "uri": "https:/data.glygen.org/ln2downloads/unicarbkb/01_11_2022/known_33806155_glygen.csv"
                }
            },
            {
                "uri": {
                    "filename": "known_33167210_glygen.csv",
                    "uri": "https:/data.glygen.org/ln2downloads/unicarbkb/01_11_2022/known_33167210_glygen.csv"
                }
            },
            {
                "uri": {
                    "filename": "known_legacy_reis_gastric_ags.csv",
                    "uri": "https:/data.glygen.org/ln2downloads/unicarbkb/01_11_2022/known_legacy_reis_gastric_ags.csv"
                }
            },
            {
                "uri": {
                    "filename": "known_legacy_sarscov2_glygen.csv",
                    "uri": "https:/data.glygen.org/ln2downloads/unicarbkb/01_11_2022/known_legacy_sarscov2_glygen.csv"
                }
            },
            {
                "uri": {
                    "filename": "unknown_20131323_glygen.csv",
                    "uri": "https:/data.glygen.org/ln2downloads/unicarbkb/01_11_2022/unknown_20131323_glygen.csv"
                }
            },
            {
                "uri": {
                    "filename": "fuzzy_simplecell_all_glygen_qc.csv",
                    "uri": "https:/data.glygen.org/ln2downloads/unicarbkb/01_11_2022/fuzzy_simplecell_all_glygen_qc.csv"
                }
            },
            {
                "uri": {
                    "filename": "mixed_34089345_glygen.csv",
                    "uri": "https:/data.glygen.org/ln2downloads/unicarbkb/01_11_2022/mixed_34089345_glygen.csv"
                }
            },
            {
                "uri": {
                    "filename": "known_33064451_glygen.csv",
                    "uri": "https:/data.glygen.org/ln2downloads/unicarbkb/01_11_2022/known_33064451_glygen.csv"
                }
            },
            {
                "uri": {
                    "filename": "unknown_33993882_glygen.csv",
                    "uri": "https:/data.glygen.org/ln2downloads/unicarbkb/01_11_2022/unknown_33993882_glygen.csv"
                }
            },
            {
                "uri": {
                    "filename": "fuzzy_legacy_human_proteoform_glycosylation_sites_unicarbkb_qin_isoforms.csv",
                    "uri": "https:/data.glygen.org/ln2downloads/unicarbkb/01_11_2022/fuzzy_legacy_human_proteoform_glycosylation_sites_unicarbkb_qin_isoforms.csv"
                }
            },
            {
                "uri": {
                    "filename": "unknown_34083726_glygen.csv",
                    "uri": "https:/data.glygen.org/ln2downloads/unicarbkb/01_11_2022/unknown_34083726_glygen.csv"
                }
            },
            {
                "uri": {
                    "filename": "known_legacy_larsson_xyl.csv",
                    "uri": "https:/data.glygen.org/ln2downloads/unicarbkb/01_11_2022/known_legacy_larsson_xyl.csv"
                }
            },
            {
                "uri": {
                    "filename": "unknown_34327564_glygen.csv",
                    "uri": "https:/data.glygen.org/ln2downloads/unicarbkb/01_11_2022/unknown_34327564_glygen.csv"
                }
            },
            {
                "uri": {
                    "filename": "fuzzy_legacy_qin_oglanac.csv",
                    "uri": "https:/data.glygen.org/ln2downloads/unicarbkb/01_11_2022/fuzzy_legacy_qin_oglanac.csv"
                }
            },
            {
                "uri": {
                    "filename": "unknown_34191522_glygen.csv",
                    "uri": "https:/data.glygen.org/ln2downloads/unicarbkb/01_11_2022/unknown_34191522_glygen.csv"
                }
            },
            {
                "uri": {
                    "filename": "n_sequon_info.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-1.12.3/misc/n_sequon_info.csv"
                }
            }
        ],
        "output_subdomain": [
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-1.12.3/reviewed/sarscov1_proteoform_glycosylation_sites_unicarbkb.csv",
                    "filename": "sarscov1_proteoform_glycosylation_sites_unicarbkb.csv"
                }
            },
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-1.12.3/reviewed/sarscov1_proteoform_glycosylation_sites_unicarbkb.stat.csv",
                    "filename": "sarscov1_proteoform_glycosylation_sites_unicarbkb.stat.csv"
                }
            }
        ]
    },
    "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"
                    },
                    {
                        "category_value": "Proteoform",
                        "category_name": "molecule"
                    },
                    {
                        "category_value": "SARS coronavirus (SARS-CoV-1)",
                        "category_name": "species"
                    },
                    {
                        "category_value": "UniCarbKB glycosylation sites",
                        "category_name": "tags"
                    },
                    {
                        "category_value": "csv",
                        "category_name": "file_type"
                    },
                    {
                        "category_value": "internal",
                        "category_name": "scope"
                    }
                ]
            }
        }
    ],
    "provenance_domain": {
        "contributors": [
            {
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "name": "Karina Martinez",
                "email": "karinamartinez@gwu.edu"
            },
            {
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "name": "Jeet Vora",
                "email": "jeetvora@gwu.edu"
            },
            {
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "name": "Robel Kahsay",
                "email": "rykahsay@email.gwu.edu"
            },
            {
                "affiliation": "Griffith University",
                "contribution": [
                    "contributedBy"
                ],
                "name": "Matthew Campbell",
                "email": "m.campbell2@griffith.edu.au"
            }
        ],
        "license": "https://creativecommons.org/licenses/by/4.0/",
        "created": "2022-03-31T15:05:49.377202",
        "modified": "2022-03-31T15:41:26.516758",
        "version": "1.12.1",
        "name": "SARS-CoV1 Glycosylation Sites (UniCarbKB)"
    },
    "description_domain": {
        "keywords": [
            "proteins",
            "canonical",
            "glycosylation sites",
            "glycan",
            "UniCarbKB",
            "GlyTouCan"
        ],
        "pipeline_steps": [
            {
                "name": "make-proteoform-dataset.py",
                "step_number": 1,
                "version": "",
                "input_list": [
                    {
                        "uri": "http://data.glygen.org/ln2downloads/unicarbkb/01_11_2022/known_33064451_glygen.csv",
                        "filename": "known_33064451_glygen.csv"
                    },
                    {
                        "uri": "https://data.glygen.org/ln2releases/v-1.12.1/reviewed/sarscov1_protein_masterlist.csv",
                        "filename": "sarscov1_protein_masterlist.csv"
                    },
                    {
                        "uri": "http://data.glygen.org/ln2downloads/glytoucan/gtc_gtc_02_04_2022/export/classification.tsv",
                        "filename": "classification.tsv"
                    },
                    {
                        "uri": "http://data.glygen.org/ln2downloads/glytoucan/gtc_gtc_02_04_2022/export/glycan_properties.tsv",
                        "filename": "glycan_properties.tsv"
                    },
                    {
                        "uri": "https://data.glygen.org/ln2releases/v-1.12.1/reviewed/sarscov1_protein_allsequences.fasta",
                        "filename": "sarscov1_protein_allsequences.fasta"
                    },
                    {
                        "uri": "https://data.glygen.org/ln2releases/v-1.12.1/misc/aadict.csv",
                        "filename": "aadict.csv"
                    },
                    {
                        "uri": "https://data.glygen.org/ln2releases/v-1.12.1/reviewed/glycan_xref_unicarbkb.csv",
                        "filename": "glycan_xref_unicarbkb.csv"
                    }
                ],
                "output_list": [
                    {
                        "uri": "https://data.glygen.org/ln2releases/v-1.12.1/reviewed/sarscov1_proteoform_glycosylation_sites_unicarbkb.csv",
                        "filename": "sarscov1_proteoform_glycosylation_sites_unicarbkb.csv"
                    }
                ],
                "prerequisite": [],
                "description": ""
            }
        ]
    },
    "execution_domain": {
        "external_data_endpoints": [],
        "environment_variables": {},
        "script_driver": "",
        "software_prerequisites": [
            {
                "version": "2.7.5",
                "name": "Python",
                "uri": {
                    "uri": "https://www.python.org/download/releases/2.7.5/",
                    "filename": ""
                }
            }
        ],
        "script": [
            {
                "uri": {
                    "uri": "https://github.com/glygener/glygen-backend-integration/blob/master/pipeline/integrator/make-proteoform-dataset.py",
                    "filename": "make-proteoform-dataset.py"
                }
            }
        ]
    },
    "parametric_domain": [
        {
            "step": "",
            "value": "",
            "param": ""
        }
    ],
    "usability_domain": [
        "This is a SARS-CoV1 Glycosylation Sites (UniCarbKB) dataset. It contains a list of SARS coronavirus (SARS-CoV-1) [taxonomy:694009] proteins with information on glycosylation sites and associated glycans (if available) from UniCarbKB. The listed proteins (UniProtKB accession) are part of the GlyGen UniProtKB SARS CoV1 Proteome Masterlist (https://data.glygen.org/GLY_000467). If you use this dataset, please provide proper attribution to UniCarbKB, EMBL-EBI-UniProtKB, and GlyGen. Cite: PMID:24234447, PMID:33237286, and PMID:31616925. This dataset can be used for obtaining a list of glycosylation sites for canonical glycoproteins from the UnicarbKB database."
    ],
    "object_id": "https://biocomputeobject.org/GLY_000628/1.12.1",
    "spec_version": "1.3",
    "etag": "68b81748216440545d31785f239a237a"
}