CIMAC-CIDC/cidc-schemas

{
  "$schema": "metaschema/strict_meta_schema.json#",
  "$id": "artifact_core",
  "title": "Artifact Core",
  "type": "object",
  "description": "Any file artifact associated with a clinical trial.",
  "inheritableBase": true,
  "properties": {
    "upload_placeholder": {
      "$comment": "It might be something like random uuid or `upload_job_id` to help trace everything back.",
      "description": "A placeholder for when artifact file is being uploaded.",
      "type": "string"
    },
    "artifact_creator": {
      "description": "The name of the center that created this artifact.",
      "type": "string",
      "enum": ["DFCI", "Mount Sinai", "Stanford", "MD Anderson"]
    },
    "uploader": {
      "description": "The name of the person uploading the artifact.",
      "type": "string"
    },
    "uuid": {
      "description": "UUID of artifact.",
      "type": "string"
    },
    "file_name": {
      "$comment": "This supposedly get generated from 'gcs_uri_format' field in template schema def.",
      "description": "The name of the file with extension.",
      "type": "string"
    },
    "object_url": {
      "description": "URL to artifact within Google Bucket.",
      "type": "string"
    },
    "uploaded_timestamp": {
      "description": "Timestamp of when artifact was loaded into the system.",
      "format": "date-time",
      "type": "string"
    },
    "file_size_bytes": {
      "description": "File size in bytes.",
      "type": "integer"
    },
    "md5_hash": {
      "description": "MD5 Hash of artifact. Not available for composite GCS objects.",
      "type": "string"
    },
    "crc32c_hash": {
      "description": "CRC32c Hash of artifact.",
      "type": "string"
    },
    "visible": {
      "description": "Indicates if the artifact is visible.  If set to false, the artifact is effectively deleted.",
      "type": "boolean"
    },
    "artifact_category": {
      "description": "Artifact category.",
      "type": "string",
      "enum": [
        "Assay Artifact from CIMAC",
        "Pipeline Artifact",
        "Manifest File"
      ]
    },
    "data_format": {
      "description": "Data Format.",
      "type": "string",
      "enum": [
        "FASTA",
        "FASTQ.GZ",
        "VCF.GZ",
        "IMAGE",
        "VCF",
        "CSV",
        "TSV",
        "XLSX",
        "NPX",
        "ELISA",
        "BAM",
        "BAM.BAI",
        "MAF",
        "BINARY",
        "TEXT",
        "ZIP",
        "FCS",
        "GZ",
        "RCC",
        "JSON",
        "YAML",
        "PDF",
        "[NOT SET]"
      ]
    },
    "facet_group": {
      "description": "The internal data category for this artifact",
      "type": "string"
    }
  },
  "anyOf": [
    {
      "required": [
        "object_url",
        "uploaded_timestamp",
        "file_size_bytes",
        "artifact_category",
        "md5_hash"
      ]
    },
    {
      "required": [
        "object_url",
        "uploaded_timestamp",
        "file_size_bytes",
        "artifact_category",
        "crc32c_hash"
      ]
    },
    {
      "required": ["upload_placeholder"]
    }
  ]
}