{
    "$schema": "http://json-schema.org/draft-06/schema#",
    "$id": "https://nextstrain.org/schemas/dataset/v2",
    "type": "object",
    "title": "Nextstrain dataset v2",
    "description": "Typically produced by Augur (`augur export v2`) and consumed by Auspice.  Combines dataset v1 meta and tree files together into one file, with additional changes.",
    "additionalProperties": false,
    "required": ["version", "meta", "tree"],
    "properties": {
        "version" : {
            "description": "Major schema version",
            "const": "v2"
        },
        "meta": {
            "type": "object",
            "$comment": "Metadata associated with phylogeny",
            "additionalProperties": false,
            "required": ["updated", "panels"],
            "properties" : {
                "title" : {
                    "description": "Auspice displays this at the top of the page",
                    "type" : "string"
                },
                "updated" : {
                    "description": "Auspice displays this (currently only in the footer)",
                    "type" : "string",
                    "pattern": "^[0-9X]{4}-[0-9X]{2}-[0-9X]{2}$"
                },
                "build_url" : {
                    "$ref": "https://nextstrain.org/schemas/auspice/config/v2#/properties/build_url"
                },
                "description" : {
                    "description": "Auspice displays this currently in the footer.",
                    "$comment": "Generally a description of the phylogeny and/or acknowledgements in Markdown format.",
                    "type": "string"
                },
                "maintainers": {
                    "$ref": "https://nextstrain.org/schemas/auspice/config/v2#/properties/maintainers"
                },
                "genome_annotations": {
                    "$ref": "https://nextstrain.org/schemas/augur/annotations"
                },
                "filters": {
                    "$ref": "https://nextstrain.org/schemas/auspice/config/v2#/properties/filters"
                },
                "panels": {
                    "$ref": "https://nextstrain.org/schemas/auspice/config/v2#/properties/panels"
                },
                "extensions": {
                    "$ref": "https://nextstrain.org/schemas/auspice/config/v2#/properties/extensions"
                },
                "geo_resolutions": {
                    "description": "The available options for the geographic resolution dropdown, and their lat/long information",
                    "type": "array",
                    "minItems": 1,
                    "items": {
                        "type": "object",
                        "description": "Each object here is an individual geo resolution",
                        "additionalProperties": false,
                        "required": ["key", "demes"],
                        "properties": {
                            "key": {
                                "type": "string",
                                "description": "Trait key - must be specified on nodes (e.g. 'country')"
                            },
                            "title": {
                              "type": "string",
                              "description": "The title to display in the geo resolution dropdown. Optional -- if not provided then `key` will be used."
                            },
                            "demes": {
                                "type": "object",
                                "description": "Mapping from deme (trait values) to lat/long",
                                "$comment": "Each value defined across the tree needs to be present here, else Auspice cannot display the deme appropriately",
                                "patternProperties": {
                                    "^[a-z_]+$": {
                                        "type": "object",
                                        "additionalProperties": false,
                                        "properties": {
                                            "latitude": {
                                                "type": "number",
                                                "minimum": -90,
                                                "maximum": 90
                                            },
                                            "longitude": {
                                                "type": "number",
                                                "minimum": -180,
                                                "maximum": 180
                                            }
                                        }
                                    }
                                }
                            }
                        }
                    }
                },
                "colorings": {
                    "type": "array",
                    "items": {
                        "allOf": [
                            {"$ref": "https://nextstrain.org/schemas/auspice/config/v2#/properties/colorings/items"}
                        ],
                        "required": ["key", "type"]
                    }
                },
                "display_defaults": {
                    "$ref": "https://nextstrain.org/schemas/auspice/config/v2#/properties/display_defaults"
                },
                "data_provenance": {
                    "$ref": "https://nextstrain.org/schemas/auspice/config/v2#/properties/data_provenance"
                }
            }
        },
        "tree": {
            "description": "One or more phylogenies using a nested JSON structure",
            "oneOf": [
                {"$ref": "#/$defs/tree"},
                {
                    "type": "array",
                    "minItems": 1,
                    "items": {"$ref": "#/$defs/tree"}
                }
            ]
        },
        "root_sequence": {
            "$ref": "https://nextstrain.org/schemas/dataset/root-sequence"
        }
    },
    "$defs": {
        "tree": {
            "type" : "object",
            "$comment": "The phylogeny in a nested JSON structure",
            "additionalProperties": false,
            "required": ["name", "node_attrs"],
            "properties": {
                "name": {
                    "description": "Strain name. Must be unique. No spaces",
                    "type": "string"
                },
                "node_attrs": {
                    "description": "attributes associated with the node (sequence, date, location) as opposed to changes from one node to another.",
                    "type": "object",
                    "anyOf": [
                      {"required": ["div"]},
                      {"required": ["num_date"]}
                    ],
                    "properties": {
                        "div": {
                            "description": "Node (phylogenetic) divergence",
                            "$comment": "Cumulative (root = 0)",
                            "type": "number"
                        },
                        "num_date": {
                            "$comment": "Sample date in decimal format (e.g. 2012.1234)",
                            "$comment": "This is the only date information. We no longer have string dates.",
                            "type": "object",
                            "required": ["value"],
                            "properties": {
                                "value": {"type": "number"},
                                "confidence": {
                                    "description": "Confidence of the node date",
                                    "type": "array",
                                    "items": [
                                        {"type": "number"},
                                        {"type": "number"}
                                    ]
                                }
                            }
                        },
                        "vaccine": {
                            "description": "Vaccine information",
                            "properties": {
                                "serum": {
                                    "description": "strain used to raise sera (for ???)",
                                    "$comment": "Currently in the flu & dengue trees",
                                    "type": "boolean"
                                },
                                "selection_date": {
                                    "description": "Vaccine selection date",
                                    "$comment": "this is currently in the metadata JSON",
                                    "type": "string",
                                    "pattern": "^[0-9X]{4}-[0-9X]{2}-[0-9X]{2}$"
                                },
                                "start_date": {
                                    "description": "Vaccine usage start date",
                                    "type": "string",
                                    "pattern": "^[0-9X]{4}-[0-9X]{2}-[0-9X]{2}$"
                                },
                                "end_date": {
                                    "description": "When the vaccine was stopped",
                                    "$comment": "if vaccine still in use, don't include this property",
                                    "type": "string",
                                    "pattern": "^[0-9X]{4}-[0-9X]{2}-[0-9X]{2}$"
                                }
                            }
                        },
                        "hidden": {
                            "$comment": "Instruct auspice to hide the branches from this node to it's children",
                            "type": "string",
                            "enum": ["always", "timetree", "divtree"]
                        },
                        "url": {
                            "description": "URL of the sequence (usually https://www.ncbi.nlm.nih.gov/nuccore/...)",
                            "$comment": "terminal nodes only",
                            "type": "string",
                            "pattern": "^https?://.+$"
                        },
                        "author": {
                            "description": "Author information (terminal nodes only)",
                            "type": "object",
                            "required": ["value"],
                            "properties": {
                                "value": {
                                    "description": "unique value for this publication. Displayed as-is by auspice.",
                                    "type": "string"
                                },
                                "title": {
                                  "description": "Publication title",
                                  "type": "string"
                                },
                                "journal": {
                                    "description": "Journal title (including year, if applicable)",
                                    "type": "string"
                                },
                                "paper_url": {
                                    "description": "URL link to paper (if available)",
                                    "type": "string",
                                    "pattern": "^https?://.+$"
                                }
                            }
                        },
                        "accession": {
                            "description": "Sequence accession number",
                            "$comment": "terminal nodes only",
                            "type": "string",
                            "pattern": "^[0-9A-Za-z-_.]+$"
                        }
                    },
                    "patternProperties": {
                        "(?!div|num_date|vaccine|hidden|url|author|accession)(^.+$)": {
                            "description": "coloring / geo resolution information attached to this node",
                            "$comment": "property name must match the `key` property provided as a coloring / geo_resolution",
                            "$comment": "JSON schema defined properties must also validate agaisnt patternProperties if they match. Hence the complex regex.",
                            "type": "object",
                            "required": ["value"],
                            "properties": {
                                "value": {
                                    "type": ["string", "number", "boolean"]
                                },
                                "confidence": {
                                    "description": "Confidence of the trait date",
                                    "$comment": "Should we use different keys for the two structures here?",
                                    "oneOf": [
                                        {
                                            "type": "array",
                                            "$comment": "Only used if the value is numeric",
                                            "items": [
                                                {"type": "number"},
                                                {"type": "number"}
                                            ]
                                        },
                                        {
                                            "type": "object",
                                            "$comment": "Only used if the value is a string",
                                            "$comment": "alternative possibilities & their confidence values",
                                            "patternProperties": {
                                                "^.+$": {
                                                    "type": "number",
                                                    "minimum": 0.0,
                                                    "maximum": 1.0
                                                }
                                            }
                                        }
                                    ]
                                },
                                "entropy": {
                                    "$comment": "Auspice uses this to control opacity of the color-by",
                                    "type": "number"
                                }
                            }
                        }
                    }
                },
                "branch_attrs": {
                    "description": "attributes associated with the branch from the parent node to this node, such as branch lengths, mutations, support values",
                    "type": "object",
                    "properties": {
                        "labels": {
                            "description": "Node labels",
                            "$comment": "Auspice scans this to generate the branch labels dropdown",
                            "patternProperties": {
                                "^[a-zA-Z0-9]+$": {
                                    "$comment": "e.g. clade->3c3a",
                                    "$comment": "string is parsed unchanged by Auspice",
                                    "type": "string"
                                }
                            }
                        },
                        "mutations": {
                            "description": "Mutations on the branch leading to this node. 1-based numbering (same as genome_annotations)",
                            "type": "object",
                            "additionalProperties": false,
                            "properties":  {
                                "nuc": {
                                    "description": "nucleotide mutations",
                                    "type": "array",
                                    "items": {
                                      "type": "string",
                                      "pattern": "^[ATCGNYRWSKMDVHB-][0-9]+[ATCGNYRWSKMDVHB-]$"
                                    }
                                }
                            },
                            "patternProperties": {
                                "^(?!nuc$)[a-zA-Z0-9*_-]+$": {
                                    "description": "Amino acid mutations for this CDS",
                                    "$comment": "properties must exist in the meta.JSON -> annotation object",
                                    "type": "array",
                                    "items": {
                                      "pattern": "^[A-Z*-][0-9]+[A-Z*-]$"
                                    }
                                }
                            }
                        }
                    }
                },
                "children": {
                    "description": "Child nodes. Recursive structure. Terminal nodes do not have this property.",
                    "$comment": "Polytomies (more than 2 items) allowed, as are nodes with a single child.",
                    "type": "array",
                    "minItems": 1,
                    "items": {"$ref": "#/$defs/tree"}
                }
            }
        }
    }
}