submission_form_schema.json

{
	"$id": "https://github.com/inab/OEB_level2_data_migration/submission-form-json-schemas",
	"$schema": "https://json-schema.org/draft/2019-09/schema",
	"title": "Schema that defines the set of fields to be filled for a submission to OpenEBench database",
	"type": "object",
	"$defs": {
		"ParticipantTool": {
			"title": "Participant Tool",
			"description": "The id of the tool used to generate the dataset. Should be selected by uploader, using API",
			"type": "string",
			"pattern": "^OEBT[0-9]{3}[A-Z0-9]{7}$",
			"options": {
				"hidden": true
			}
		},
		"DataVersion": {
			"title": "Version",
			"description": "Version (or release date) of the dataset",
			"minLength": 1,
			"type": ["string","number"],
			"options": {
				"hidden": true
			 }
		},
		"DataContacts": {
			"title": "Contacts",
			"description": "Emails of the dataset contact(s). Should be registered in Mongo and OIDC, as they are going to be used to do the proper internal mappings",
			"type": "array",
			"minItems": 1,
			"uniqueItems": true,
			"items": {
				"type": "string",
				"format": "autocomplete",
				"options": {
					"autocomplete": {
						"search": "search_zza",
						"getResultValue": "getResultValue_zza",
						"renderResult": "renderResult_zza",
						"autoSelect": true
					}
				}
			}
		},
		"ParticipantFile": {
			"title": "Participant file associated",
			"description": "Path or URI (e.g. DOI) to the participant file associated with the consolidated results",
			"type": "string",
			"readonly": true
		},
		"ParticipantElements": {
			"type": "object",
			"properties": {
				"tool_id": {
					"$ref": "#/$defs/ParticipantTool"
				},
				"data_version": {
					"$ref": "#/$defs/DataVersion"
				},
				"data_contacts": {
					"$ref": "#/$defs/DataContacts"
				},
				"participant_id": {
					"type": "string",
					"title": "The id / name of the tool which generated this dataset",
					"minLength": 1,
					"primary_key": {
						"name": "participant_id_key",
						"members": [ "." ]
					}
				},
				"exclude": {
					"title": "Should participant be excluded?",
					"type": "boolean",
					"default": false
				},
				"participant_file": {
					"$ref": "#/$defs/ParticipantFile"
				}
			},
			"required": ["tool_id","data_version","data_contacts"]
		}
	},
	"properties": {
		"consolidated_oeb_data": {
			"title": "Consolidated dataset path or URL",
			"description": "Path or publicly reachable URL to the consolidated dataset, coming from an OpenEBench standardized benchmarking workflow (e.g.https://github.com/inab/TCGA_benchmarking_workflow",
			"type": "string",
			"readonly": true,
			"options": {
			    "hidden": true
			 }
		},
		"data_visibility": {
			"title": "Datasets visibility",
			"description": "The desired visibility of the submitted datasets, which must be acknowledged by the APIs",
			"type": "string",
			"enum": ["public", "community", "challenge", "participant"],
			"default": "public",
			"readonly": true
		},
		
		"benchmarking_event_id": {
			"title": "Benchmarking event",
			"description": "The unique id of the benchmarking event the dataset belongs to, as is returned by the API",
			"type": "string",
			"pattern": "^OEBE[0-9]{3}[A-Z0-9]{7}$",
			"readonly": true
		},
		"participant_file": {
			"$ref": "#/$defs/ParticipantFile"
		},
		"community_id": {
			"title": "OEB community",
			"description": "The unique id of the community where the data should be uploaded. Should come from VRE workflow. (This parameter is not needed any more)",
			"type": "string",
			"pattern": "^OEBC[0-9]{3}$",
			"options": {
			    "hidden": true
			 }
		},
		"tool_id": {
			"$ref": "#/$defs/ParticipantTool"
		},
		"tool_mapping": {
			"title": "List of tools",
			"type": "array",
			"minItems": 1,
			"items": {
				"$ref": "#/$defs/ParticipantElements"
			},
			"unevaluatedProperties": false
		},
		"tool_selection": {
			"title": "Participant Tool",
			"description": "The id of the tool used to generate the dataset. Should be selected by uploader, using API",
			"type": "string",
			"format": "selectize",
			"options": {
			    "selectize": {
					"load": "load",
					"render": "render",
					"placeholder": "Name of your tool",
					"valueField": "_id",
					"labelField": "name",
					"searchField": "name",
					"options": []
			    }
			}
		},

		"dataset_submission_id": {
			"title": "Dataset submission id",
			"description": "The unique identifier of the dataset which will hold the identifiers of all the loaded/updated datasets. Beware, this id should be unique! If it is not set, a UUIDv1 is used",
			"type": "string",
			"options": {
			    "hidden": true
			 }
		},
		"workflow_oeb_id": {
			"title": "OEB Workflow",
			"description": "The id of the workflow (as a tool) used to compute the assessment metrics. Should be associated to VRE tool",
			"type": "string",
			"pattern": "^OEBT[0-9]{3}[A-Z0-9]{7}$",
			"options": {
			    "hidden": true
			 }
		},
		"data_model_repo": {
			"title": "OpenEBench benchmarking data model repository",
			"description": "The git repository URI, needed to check out the data model",
			"type": "string",
			"format": "uri",
			"minLength": 1,
			"default": "https://github.com/inab/benchmarking-data-model.git",
			"options": {
			    "hidden": true
			 }
		},
		"data_model_tag": {
			"title": "Data model tag",
			"description": "Either the tag, branch or checkout hash needed to fetch the right version of the data model",
			"type": "string",
			"minLength": 1,
			"default": "6495b7317f830ad739591be2de1e279ea6c4c0d8",
			"options": {
			    "hidden": true
			 }
		},
		"data_model_reldir": {
			"title": "Data model relative path",
			"description": "Within a checked-out copy of the benchmarking data model repository, the relative path to the data model",
			"type": "string",
			"format": "uri-reference",
			"default": "json-schemas/1.0.x",
			"options": {
			    "hidden": true
			 }
		},
		"fix_original_ids": {
			"title": "Fix non-conformant original ids",
			"description": "Fix original ids provided by the minimal datasets, so they conform the original id naming conventions",
			"type": "boolean",
			"default": true
		},
		"data_storage_endpoint": {
			"$comment": "This key is only allowed for backward compatibility",
			"title": "Storage endpoint",
			"description": "Storage endpoint used to submit the assessment dataset to a perdurable storage which is able to provide a persistent id (e.g. a DOI)",
			"type": "string",
			"format": "uri",
			"minLength": 1
		}
	},
	"dependentSchemas": {
		"tool_id": {
			"$ref": "#/$defs/ParticipantElements"
		}
	},
	"oneOf": [
		{
			"required": ["tool_id", "participant_file"],
			"not": {
				"required": ["tool_mapping"]
			}
		},
		{
			"required": ["tool_mapping"],
			"not": {
				"required": ["tool_id"]
			}
		}
	],
	"required": [
		"consolidated_oeb_data",
		"data_visibility",
		"benchmarking_event_id"
	]
}