{
    "io_domain": {
        "input_subdomain": [
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2data/releases/data/v-1.8.24/source/mgi_orthologs_hom_allorganism_release6.14_060419_12-33pm.csv",
                    "filename": "mgi_orthologs_hom_allorganism_release6.14_060419_12-33pm.csv"
                }
            },
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2data/releases/data/v-1.8.24/source/mouse_human_orthologs_oma_release-24_061319_03-45pm.csv",
                    "filename": "mouse_human_orthologs_oma_release-24_061319_03-45pm.csv"
                }
            },
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2data/releases/data/v-1.8.24/source/rat_human_orthologs_oma_release-24_061319_12-24pm.csv",
                    "filename": "rat_human_orthologs_oma_release-24_061319_12-24pm.csv"
                }
            },
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2data/releases/data/v-1.8.24/source/mgi_homologs.csv",
                    "filename": "mgi_homologs.csv"
                }
            },
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2data/releases/data/v-1.8.24/reviewed/mouse_protein_xref_mgi.csv",
                    "filename": "mouse_protein_xref_mgi.csv"
                }
            },
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2data/releases/data/v-1.8.24/reviewed/human_protein_xref_hgnc.csv",
                    "filename": "human_protein_xref_hgnc.csv"
                }
            },
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2data/releases/data/v-1.8.24/reviewed/human_protein_xref_oma.csv",
                    "filename": "human_protein_xref_oma.csv"
                }
            },
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2data/releases/data/v-1.8.24/reviewed/mouse_protein_xref_oma.csv",
                    "filename": "mouse_protein_xref_oma.csv"
                }
            },
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2data/releases/data/v-1.8.24/reviewed/mouse_protein_xref_geneid.csv",
                    "filename": "mouse_protein_xref_geneid.csv"
                }
            },
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2data/releases/data/v-1.8.24/reviewed/rat_protein_xref_geneid.csv",
                    "filename": "rat_protein_xref_geneid.csv"
                }
            },
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2data/releases/data/v-1.8.24/reviewed/human_protein_xref_geneid.csv",
                    "filename": "human_protein_xref_geneid.csv"
                }
            },
            {
                "uri": {
                    "uri": "https://data.glygen.org/ln2data/releases/data/v-1.8.24/reviewed/rat_protein_xref_oma.csv",
                    "filename": "rat_protein_xref_oma.csv"
                }
            }
        ],
        "output_subdomain": [
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "http://data.glygen.org/ln2data/releases/data/v-1.8.24/reviewed/protein_homolog_clusters.csv",
                    "filename": "protein_homolog_clusters.csv"
                }
            },
            {
                "mediatype": "csv",
                "uri": {
                    "uri": "http://data.glygen.org/ln2data/releases/data/v-1.8.24/reviewed/protein_homolog_clusters.stat.csv",
                    "filename": "protein_homolog_clusters.stat.csv"
                }
            }
        ]
    },
    "spec_version": "1.3.0",
    "object_id": "https://biocomputeobject.org/GLY_000258/v-1.8.24",
    "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"
            }
        ],
        "created": "2019-09-06T14:06:38.985197",
        "modified": "2020-04-29T23:01:49.049875",
        "version": "v-1.8.24"
    },
    "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-1.8.24/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-1.8.24/source/mgi_homologs.csv",
                        "filename": "mgi_homologs.csv"
                    },
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-1.8.24/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-1.8.24/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-1.8.24/reviewed/mouse_protein_xref_mgi.csv",
                        "filename": "mouse_protein_xref_mgi.csv"
                    },
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-1.8.24/reviewed/human_protein_xref_hgnc.csv",
                        "filename": "human_protein_xref_hgnc.csv"
                    },
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-1.8.24/reviewed/human_protein_xref_oma.csv",
                        "filename": "human_protein_xref_oma.csv"
                    },
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-1.8.24/reviewed/mouse_protein_xref_oma.csv",
                        "filename": "mouse_protein_xref_oma.csv"
                    },
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-1.8.24/reviewed/mouse_protein_xref_geneid.csv",
                        "filename": "mouse_protein_xref_geneid.csv"
                    },
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-1.8.24/reviewed/rat_protein_xref_geneid.csv",
                        "filename": "rat_protein_xref_geneid.csv"
                    },
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-1.8.24/reviewed/human_protein_xref_geneid.csv",
                        "filename": "human_protein_xref_geneid.csv"
                    },
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-1.8.24/reviewed/rat_protein_xref_oma.csv",
                        "filename": "rat_protein_xref_oma.csv"
                    }
                ],
                "output_list": [
                    {
                        "uri": "https://data.glygen.org/ln2data/releases/data/v-1.8.24/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/GW-HIVE/glygen-backend-integration/blob/master/pipeline/integrator/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": "42c8ea80b64611a46a742fed4f9a391f",
    "_id": {
        "$oid": "5d72a02ef5c47b0f43423373"
    }
}