references/intake-schema.json

{
  "$schema": "https://json-schema.org/draft/2020-12/schema",
  "title": "NGS analysis intake",
  "type": "object",
  "additionalProperties": false,
  "required": ["input_type", "assay", "objective"],
  "properties": {
    "input_type": {
      "type": "string",
      "enum": ["bcl_run_folder", "fastq", "bam_or_cram", "count_matrix", "vcf", "unknown"]
    },
    "assay": {
      "type": "string",
      "enum": [
        "wgs",
        "wes",
        "targeted_panel",
        "bulk_rnaseq",
        "scrnaseq",
        "snrnaseq",
        "atacseq",
        "chipseq",
        "cutandrun",
        "cutandtag",
        "amplicon_microbiome",
        "shotgun_metagenomics",
        "fastq_qc_only",
        "unknown"
      ]
    },
    "objective": {
      "type": "array",
      "items": {
        "type": "string",
        "enum": [
          "demultiplex",
          "qc",
          "trim",
          "align",
          "variant_calling",
          "germline_variant_calling",
          "somatic_variant_calling",
          "umi_panel_variant_calling",
          "variant_annotation",
          "expression_counts",
          "differential_expression",
          "cell_matrix",
          "cell_qc",
          "cell_annotation",
          "post_count_qc",
          "peak_calling",
          "differential_accessibility",
          "differential_binding",
          "taxonomic_profile",
          "functional_profile",
          "report_only"
        ]
      },
      "minItems": 1,
      "uniqueItems": true
    },
    "organism": {"type": "string"},
    "reference": {
      "type": "object",
      "additionalProperties": true,
      "properties": {
        "genome": {"type": "string"},
        "fasta": {"type": "string"},
        "gtf": {"type": "string"},
        "bed": {"type": "string"},
        "known_sites": {"type": "array", "items": {"type": "string"}}
      }
    },
    "library": {
      "type": "object",
      "additionalProperties": true,
      "properties": {
        "paired_end": {"type": "boolean"},
        "strandedness": {
          "type": "string",
          "enum": ["forward", "reverse", "unstranded", "unknown"]
        },
        "umi": {"type": "boolean"},
        "chemistry": {"type": "string"},
        "read_length": {"type": "string"},
        "adapters_or_primers": {"type": "array", "items": {"type": "string"}}
      }
    },
    "study_design": {
      "type": "object",
      "additionalProperties": true,
      "properties": {
        "samplesheet": {"type": "string"},
        "contrasts": {"type": "array", "items": {"type": "string"}},
        "tumor_normal": {"type": "boolean"},
        "tumor_only": {"type": "boolean"},
        "trio": {"type": "boolean"},
        "sample_model": {
          "type": "string",
          "enum": ["singleton", "cohort", "duo", "trio", "family", "tumor_only", "tumor_normal", "umi_panel", "unknown"]
        },
        "replicates": {"type": "boolean"},
        "batches": {"type": "array", "items": {"type": "string"}}
      }
    },
    "runtime": {
      "type": "object",
      "additionalProperties": true,
      "properties": {
        "executor": {
          "type": "string",
          "enum": ["local", "slurm", "sge", "aws", "gcp", "azure", "unknown"]
        },
        "container_engine": {
          "type": "string",
          "enum": ["docker", "singularity", "apptainer", "none", "unknown"]
        },
        "can_install": {"type": "boolean"},
        "allow_network": {"type": "boolean"},
        "allow_cloud_upload": {"type": "boolean"},
        "gpu_available": {"type": "boolean"}
      }
    },
    "routing_decision": {
      "type": "object",
      "additionalProperties": true,
      "properties": {
        "pipeline_key": {"type": "string"},
        "pipeline_name": {"type": "string"},
        "confidence": {"type": "string", "enum": ["high", "medium", "low"]},
        "missing_essentials": {"type": "array", "items": {"type": "string"}}
      }
    }
  }
}