{
    "object_id": "https://biocomputeobject.org/GLY_000258/v-2.0.2",
    "spec_version": "1.3.0",
    "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": "",
                        "category_name": "species"
                    },
                    {
                        "category_value": "Protein",
                        "category_name": "molecule"
                    },
                    {
                        "category_value": "Homologs",
                        "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": "Homolog Clusters",
        "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": "Karina Martinez",
                "email": "karinamartinez@gwu.edu"
            },
            {
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "name": "Amanda Bell",
                "email": "amandab2140@gwu.edu"
            },
            {
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "name": "Jonathon Keeney",
                "email": "keeneyjg@gwu.edu"
            },
            {
                "affiliation": "The George Washington University",
                "contribution": [
                    "createdBy"
                ],
                "name": "Millicent Quartey",
                "email": "mquartey@email.gwu.edu"
            },
            {
                "affiliation": "",
                "contribution": [
                    "contributedBy"
                ],
                "name": "Not Applicable",
                "email": ""
            }
        ],
        "created": "2019-09-06T14:06:38.985197",
        "modified": "2022-01-19T13:37:05.369499",
        "version": "v-2.0.2"
    },
    "description_domain": {
        "keywords": [
            "homologs",
            "orthologs",
            "MGI",
            "OMA"
        ],
        "pipeline_steps": [
            {
                "name": "make-species-agnostic-dataset.py",
                "step_number": 1,
                "version": "",
                "input_list": [
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-2.0.2/source/mouse_human_orthologs_oma_release-24_061319_03-45pm.csv",
                        "filename": "mouse_human_orthologs_oma_release-24_061319_03-45pm.csv"
                    },
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-2.0.2/source/mgi_homologs.csv",
                        "filename": "mgi_homologs.csv"
                    },
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-2.0.2/source/mouse_human_orthologs_oma_release-24_061319_03-45pm.csv",
                        "filename": "mouse_human_orthologs_oma_release-24_061319_03-45pm.csv"
                    },
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-2.0.2/source/rat_human_orthologs_oma_release-24_061319_03-45pm.csv",
                        "filename": "rat_human_orthologs_oma_release-24_061319_03-45pm.csv"
                    },
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-2.0.2/reviewed/mouse_protein_xref_mgi.csv",
                        "filename": "mouse_protein_xref_mgi.csv"
                    },
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-2.0.2/reviewed/human_protein_xref_hgnc.csv",
                        "filename": "human_protein_xref_hgnc.csv"
                    },
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-2.0.2/reviewed/human_protein_xref_oma.csv",
                        "filename": "human_protein_xref_oma.csv"
                    },
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-2.0.2/reviewed/mouse_protein_xref_oma.csv",
                        "filename": "mouse_protein_xref_oma.csv"
                    },
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-2.0.2/reviewed/mouse_protein_xref_geneid.csv",
                        "filename": "mouse_protein_xref_geneid.csv"
                    },
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-2.0.2/reviewed/rat_protein_xref_geneid.csv",
                        "filename": "rat_protein_xref_geneid.csv"
                    },
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-2.0.2/reviewed/human_protein_xref_geneid.csv",
                        "filename": "human_protein_xref_geneid.csv"
                    },
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-2.0.2/reviewed/rat_protein_xref_oma.csv",
                        "filename": "rat_protein_xref_oma.csv"
                    }
                ],
                "output_list": [
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-2.0.2/reviewed/homolog_clusters.csv",
                        "filename": "homolog_clusters.csv"
                    }
                ],
                "prerequisite": [],
                "description": "The script extracts the human, mouse and rat homologs from the input files and using xref files, maps the identifiers to UniProtKB canonical accessions. "
            }
        ]
    },
    "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-integration/blob/master/dataset-maker/make-species-agnostic-dataset.py",
                    "filename": "make-species-agnostic-dataset.py"
                }
            }
        ]
    },
    "usability_domain": [
        "The Homolog Clusters dataset contains homologs of human [taxid: 9606], mouse [taxid:10090] and rat [taxid:10116] species mapped to UniProtKB canonical accessions. The homologs data is downloaded from OMA browser and MGI. If you use this dataset please provide proper attribution to MGI, OMA Browser and GlyGen. "
    ],
    "etag": "a11bd86571c295a0a78d94966e0f73e1",
    "io_domain": {
        "input_subdomain": [
            {
                "uri": {
                    "filename": "protein_homolog_clusters.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.0.2/compiled/protein_homolog_clusters.csv"
                }
            },
            {
                "uri": {
                    "filename": "fruitfly_protein_transcriptlocus.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.0.2/reviewed/fruitfly_protein_transcriptlocus.csv"
                }
            },
            {
                "uri": {
                    "filename": "fruitfly_protein_xref_geneid.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.0.2/reviewed/fruitfly_protein_xref_geneid.csv"
                }
            },
            {
                "uri": {
                    "filename": "fruitfly_protein_xref_oma.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.0.2/reviewed/fruitfly_protein_xref_oma.csv"
                }
            },
            {
                "uri": {
                    "filename": "human_protein_transcriptlocus.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.0.2/reviewed/human_protein_transcriptlocus.csv"
                }
            },
            {
                "uri": {
                    "filename": "human_protein_xref_geneid.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.0.2/reviewed/human_protein_xref_geneid.csv"
                }
            },
            {
                "uri": {
                    "filename": "human_protein_xref_hgnc.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.0.2/reviewed/human_protein_xref_hgnc.csv"
                }
            },
            {
                "uri": {
                    "filename": "human_protein_xref_oma.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.0.2/reviewed/human_protein_xref_oma.csv"
                }
            },
            {
                "uri": {
                    "filename": "mouse_protein_transcriptlocus.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.0.2/reviewed/mouse_protein_transcriptlocus.csv"
                }
            },
            {
                "uri": {
                    "filename": "mouse_protein_xref_geneid.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.0.2/reviewed/mouse_protein_xref_geneid.csv"
                }
            },
            {
                "uri": {
                    "filename": "mouse_protein_xref_mgi.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.0.2/reviewed/mouse_protein_xref_mgi.csv"
                }
            },
            {
                "uri": {
                    "filename": "mouse_protein_xref_oma.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.0.2/reviewed/mouse_protein_xref_oma.csv"
                }
            },
            {
                "uri": {
                    "filename": "rat_protein_transcriptlocus.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.0.2/reviewed/rat_protein_transcriptlocus.csv"
                }
            },
            {
                "uri": {
                    "filename": "rat_protein_xref_geneid.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.0.2/reviewed/rat_protein_xref_geneid.csv"
                }
            },
            {
                "uri": {
                    "filename": "rat_protein_xref_oma.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.0.2/reviewed/rat_protein_xref_oma.csv"
                }
            },
            {
                "uri": {
                    "filename": "sarscov1_protein_xref_geneid.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.0.2/reviewed/sarscov1_protein_xref_geneid.csv"
                }
            },
            {
                "uri": {
                    "filename": "sarscov2_protein_xref_geneid.csv",
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.0.2/reviewed/sarscov2_protein_xref_geneid.csv"
                }
            },
            {
                "uri": {
                    "filename": "mgi_homologs.csv",
                    "uri": "https:/data.glygen.org/ln2downloads/mgi/2022_07_07/mgi_homologs.csv"
                }
            },
            {
                "uri": {
                    "filename": "fruitfly_human_oma_orthologs_july2022.csv",
                    "uri": "https:/data.glygen.org/ln2downloads/oma/2022_07_07/fruitfly_human_oma_orthologs_july2022.csv"
                }
            },
            {
                "uri": {
                    "filename": "fruitfly_mouse_oma_orthologs_july2022.csv",
                    "uri": "https:/data.glygen.org/ln2downloads/oma/2022_07_07/fruitfly_mouse_oma_orthologs_july2022.csv"
                }
            },
            {
                "uri": {
                    "filename": "fruitfly_rat_oma_orthologs_july2022.csv",
                    "uri": "https:/data.glygen.org/ln2downloads/oma/2022_07_07/fruitfly_rat_oma_orthologs_july2022.csv"
                }
            },
            {
                "uri": {
                    "filename": "human_fruitfly_oma_orthologs_july2022.csv",
                    "uri": "https:/data.glygen.org/ln2downloads/oma/2022_07_07/human_fruitfly_oma_orthologs_july2022.csv"
                }
            },
            {
                "uri": {
                    "filename": "human_mouse_oma_orthologs_july2022.csv",
                    "uri": "https:/data.glygen.org/ln2downloads/oma/2022_07_07/human_mouse_oma_orthologs_july2022.csv"
                }
            },
            {
                "uri": {
                    "filename": "human_rat_oma_orthologs_july2022.csv",
                    "uri": "https:/data.glygen.org/ln2downloads/oma/2022_07_07/human_rat_oma_orthologs_july2022.csv"
                }
            },
            {
                "uri": {
                    "filename": "mouse_fruitfly_oma_orthologs_july2022.csv",
                    "uri": "https:/data.glygen.org/ln2downloads/oma/2022_07_07/mouse_fruitfly_oma_orthologs_july2022.csv"
                }
            },
            {
                "uri": {
                    "filename": "mouse_human_oma_orthologs_july2022.csv",
                    "uri": "https:/data.glygen.org/ln2downloads/oma/2022_07_07/mouse_human_oma_orthologs_july2022.csv"
                }
            },
            {
                "uri": {
                    "filename": "mouse_rat_oma_orthologs_july2022.csv",
                    "uri": "https:/data.glygen.org/ln2downloads/oma/2022_07_07/mouse_rat_oma_orthologs_july2022.csv"
                }
            },
            {
                "uri": {
                    "filename": "rat_fruitfly_oma_orthologs_july2022.csv",
                    "uri": "https:/data.glygen.org/ln2downloads/oma/2022_07_07/rat_fruitfly_oma_orthologs_july2022.csv"
                }
            },
            {
                "uri": {
                    "filename": "rat_human_oma_orthologs_july2022.csv",
                    "uri": "https:/data.glygen.org/ln2downloads/oma/2022_07_07/rat_human_oma_orthologs_july2022.csv"
                }
            },
            {
                "uri": {
                    "filename": "rat_mouse_oma_orthologs_july2022.csv",
                    "uri": "https:/data.glygen.org/ln2downloads/oma/2022_07_07/rat_mouse_oma_orthologs_july2022.csv"
                }
            }
        ],
        "output_subdomain": [
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.0.2/reviewed/protein_homolog_clusters.csv",
                    "filename": "protein_homolog_clusters.csv"
                }
            },
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "https:/data.glygen.org/ln2releases/data/v-2.0.2/reviewed/protein_homolog_clusters.stat.csv",
                    "filename": "protein_homolog_clusters.stat.csv"
                }
            }
        ]
    }
}