#!/usr/bin/env cwl-runner
cwlVersion: v1.2
class: Workflow
id: structure_predict
label: "Core 1: Structure prediction (ESMFold / Boltz-2)"
doc: |
  Catalog entry `structure_predict` (Model surface, biolm-protocols catalog.json).
  Predicts 3D structure for a set of sequences with a single BioLM structure
  model. No protocol orchestration needed - this is a direct
  `biolm model run` call.

requirements:
  InlineJavascriptRequirement: {}
  StepInputExpressionRequirement: {}

inputs:
  sequences_file:
    type: File
    doc: "FASTA (or CSV/JSON) file of sequences to fold. See examples/structure_predict.job.yml."
  model_name:
    type: string
    default: esmfold
    doc: "Structure model slug: esmfold (primary) or boltz-2 (optional, per catalog.json)."
  backend:
    type: string
    default: platform
    doc: "'platform' (biolm.ai) or 'hub' (route via hub_url / BIOLM_BASE_API_URL)."
  hub_url:
    type: string?
    doc: "Hub gateway base URL, required when backend=hub."
  biolm_token:
    type: string?
    doc: "BIOLM_TOKEN; omit to use the ambient environment or ~/.biolm/credentials."
  demo:
    type: boolean
    default: false
    doc: "Cheap smoke-test mode: caps batch_size to 1 for a fast, low-cost run."

steps:
  fold:
    run: ../tools/biolm-model-run.cwl
    in:
      backend: backend
      hub_url: hub_url
      biolm_token: biolm_token
      model_name: model_name
      action:
        valueFrom: "predict"
      input_file: sequences_file
      batch_size:
        source: demo
        valueFrom: "$(self ? 1 : null)"
      progress:
        valueFrom: $(false)
    out: [result]

outputs:
  structures:
    type: File
    outputSource: fold/result
    doc: "JSON array of {pdb, mean_plddt, ...} records, one per input sequence."