Boltz Workflow: Batch Ligand Ranking

{
  "io": {
    "inputs": [
      {
        "ui": {
          "rows": 6,
          "control": "textarea"
        },
        "name": "protein_sequence",
        "label": "ABL1 Kinase Sequence",
        "format": "sequence",
        "maps_to": "ligand_library_loader.protein_sequence",
        "required": false,
        "value_type": "string",
        "description": "Amino-acid sequence for the shared ABL1 kinase-domain target; maps directly to the Boltz-2 protein input for each ligand run."
      },
      {
        "ui": {
          "control": "file"
        },
        "file": {
          "accept": [
            ".csv",
            "text/csv"
          ]
        },
        "name": "ligand_csv",
        "label": "Candidate Ligand CSV",
        "format": "csv",
        "maps_to": "ligand_library_loader.ligand_csv",
        "required": false,
        "value_type": "file",
        "description": "CSV text with candidate ligand names and SMILES strings; each row is passed to Boltz-2 as a separate ligand input."
      },
      {
        "ui": {
          "control": "file"
        },
        "file": {
          "accept": [
            ".a3m",
            ".fasta",
            ".fa"
          ]
        },
        "name": "msa_path",
        "label": "Precomputed MSA Path",
        "format": "a3m",
        "maps_to": "ligand_library_loader.msa_path",
        "required": false,
        "value_type": "file",
        "description": "Optional path to a precomputed MSA file; leave unset when the configured Boltz MSA server is used."
      },
      {
        "name": "use_msa_server",
        "label": "Use MSA Server",
        "default": true,
        "maps_to": "ligand_library_loader.run_options.use_msa_server",
        "value_type": "boolean",
        "description": "Use Boltz's MSA server instead of requiring a precomputed MSA file."
      },
      {
        "name": "sampling_steps",
        "label": "Sampling Steps",
        "default": 200,
        "maps_to": "ligand_library_loader.run_options.sampling_steps",
        "value_type": "integer",
        "description": "Number of Boltz sampling steps for structure generation."
      },
      {
        "name": "recycling_steps",
        "label": "Recycling Steps",
        "default": 3,
        "maps_to": "ligand_library_loader.run_options.recycling_steps",
        "value_type": "integer",
        "description": "Number of structure recycling steps."
      },
      {
        "name": "diffusion_samples",
        "label": "Diffusion Samples",
        "default": 1,
        "maps_to": "ligand_library_loader.run_options.diffusion_samples",
        "value_type": "integer",
        "description": "Number of Boltz diffusion samples to generate."
      },
      {
        "name": "output_format",
        "label": "Output Format",
        "default": "mmcif",
        "maps_to": "ligand_library_loader.run_options.output_format",
        "value_type": "string",
        "description": "Structure artifact format emitted by Boltz.",
        "allowed_values": [
          "mmcif",
          "pdb"
        ]
      },
      {
        "ui": {
          "rows": 6,
          "control": "json"
        },
        "name": "run_options",
        "label": "Boltz Batch Run Options",
        "format": "json",
        "maps_to": "ligand_library_loader.run_options",
        "advanced": true,
        "required": false,
        "value_type": "record",
        "description": "Optional structured controls for the batch workflow, including source metadata, sampling settings, and interpretation scope."
      }
    ],
    "outputs": [
      {
        "name": "scenario_context",
        "label": "Workflow Scenario Context",
        "maps_to": "batch_target_context.scenario_context",
        "description": "Source-backed target, ligand, use-case, provenance, and caveat context for this Boltz workflow."
      },
      {
        "name": "assembled_boltz_request",
        "label": "Assembled Boltz Request",
        "maps_to": "ligand_library_loader.assembled_boltz_request",
        "description": "Traceable summary of the protein, ligand, MSA, and run options prepared for Boltz."
      },
      {
        "name": "batch_summary",
        "label": "Ranked Ligand Summary",
        "maps_to": "boltz_batch_ligand_ranker.batch_summary",
        "description": "Ranked ligand table with binder probability, affinity-like value, confidence, status, and caveat flags."
      },
      {
        "name": "affinity_summary",
        "label": "Top Ligand Affinity-Style Summary",
        "maps_to": "boltz_batch_ligand_ranker.affinity_summary",
        "description": "Parsed Boltz-2 binder probability and affinity-like outputs for the top-ranked completed ligand."
      },
      {
        "name": "confidence_summary",
        "label": "Top Ligand Confidence Summary",
        "maps_to": "boltz_batch_ligand_ranker.confidence_summary",
        "description": "Parsed Boltz-2 confidence outputs for the top-ranked predicted complex."
      },
      {
        "name": "structure_artifacts",
        "label": "Top Ligand Structure Artifacts",
        "maps_to": "boltz_batch_ligand_ranker.structure_artifacts",
        "description": "File-backed predicted complex structures for the top-ranked completed ligand."
      },
      {
        "name": "run_metadata",
        "label": "Boltz Batch Run Metadata",
        "maps_to": "boltz_batch_ligand_ranker.run_metadata",
        "description": "Runtime status, per-ligand statuses, command metadata, logs, and caveats for the latest batch invocation."
      },
      {
        "name": "ranking_evidence",
        "label": "Conservative Ranking Evidence",
        "maps_to": "ranking_interpreter.prediction_evidence",
        "description": "Interpreted Boltz output evidence with request traceability and explicit scientific caveats."
      }
    ]
  },
  "tags": [
    "boltz-workflow",
    "boltz",
    "protein-ligand",
    "affinity",
    "structural-biology",
    "guided-workflow",
    "batch-ranking",
    "gpu"
  ],
  "title": "Boltz Workflow: Batch Ligand Ranking",
  "models": [
    {
      "path": "models/context",
      "alias": "batch_target_context",
      "parameters": {
        "scenario": {
          "caveat": "Boltz-2 outputs are computational structure and affinity-style predictions for hypothesis generation; they are not experimental binding, potency, selectivity, clinical, or efficacy evidence.",
          "source_pdb": "2HYY",
          "ligand_role": "candidate ligand set",
          "target_name": "Human ABL1 kinase domain",
          "disease_area": "small ligand ranking",
          "target_family": "ABL1 kinase domain",
          "workflow_name": "Batch Ligand Ranking",
          "workflow_context": "Guided Boltz-2 batch ligand ranking workflow using ABL1 kinase-domain examples",
          "workflow_question": "How does a small ligand library rank against the shared ABL1 kinase-domain target?",
          "interpretation_scope": "Learning, small-set ranking, and early hypothesis generation only",
          "protein_sequence_length": 273
        },
        "integration_step": 0.01
      }
    },
    {
      "path": "models/input_assembler",
      "alias": "ligand_library_loader",
      "parameters": {
        "workflow_kind": "batch",
        "workflow_name": "Batch Ligand Ranking",
        "integration_step": 0.01,
        "default_ligand_csv": "name,smiles,source\nImatinib,CC1=C(C=C(C=C1)NC(=O)C2=CC=C(C=C2)CN3CCN(CC3)C)NC4=NC=CC(=N4)C5=CN=CC=C5,PubChem CID 5291\nDasatinib,CC1=C(C(=CC=C1)Cl)NC(=O)C2=CN=C(S2)NC3=CC(=NC(=N3)C)N4CCN(CC4)CCO,PubChem CID 3062316\nNilotinib,CC1=C(C=C(C=C1)C(=O)NC2=CC(=CC(=C2)C(F)(F)F)N3C=C(N=C3)C)NC4=NC=CC(=N4)C5=CN=CC=C5,PubChem CID 644241\n",
        "default_run_options": {
          "source_pdb": "2HYY",
          "target_name": "Human ABL1 kinase domain",
          "workflow_name": "Batch Ligand Ranking",
          "ligand_examples": [
            "Imatinib",
            "Dasatinib",
            "Nilotinib"
          ],
          "workflow_context": "Guided Boltz-2 batch ligand ranking workflow using ABL1 kinase-domain examples",
          "interpretation_scope": "Learning, small-set ranking, and early hypothesis generation only"
        },
        "default_protein_sequence": "VSPNYDKWEMERTDITMKHKLGGGQYGEVYEGVWKKYSLTVAVKTLKEDTMEVEEFLKEAAVMKEIKHPNLVQLLGVCTREPPFYIITEFMTYGNLLDYLRECNRQEVNAVVLLYMATQISSAMEYLEKKNFIHRDLAARNCLVGENHLVKVADFGLSRLMTGDTYTAHAGAKFPIKWTAPESLAYNKFSIKSDVWAFGVLLWEIATYGMSPYPGIDLSQVYELLEKDYRMERPEGCPEKVYELMRACWQWNPSDRPSFAEIHQAFETMFQES"
      }
    },
    {
      "path": "models/core",
      "alias": "boltz_batch_ligand_ranker",
      "parameters": {
        "override": true,
        "accelerator": "gpu",
        "max_ligands": 3,
        "runtime_mode": "managed",
        "output_format": "mmcif",
        "sampling_steps": 200,
        "use_msa_server": true,
        "recycling_steps": 3,
        "diffusion_samples": 1
      }
    },
    {
      "path": "models/interpreter",
      "alias": "ranking_interpreter",
      "parameters": {
        "mode": "batch",
        "caveat": "Boltz-2 outputs are computational structure and affinity-style predictions for hypothesis generation; they are not experimental binding, potency, selectivity, clinical, or efficacy evidence.",
        "core_alias": "boltz_batch_ligand_ranker",
        "workflow_name": "Batch Ligand Ranking",
        "integration_step": 0.01
      }
    },
    {
      "path": "models/visualisation",
      "alias": "visualisation",
      "parameters": {
        "mode": "boltz_batch",
        "lab_title": "Boltz Workflow: Batch Ligand Ranking",
        "source_alias": "boltz_batch_ligand_ranker",
        "context_alias": "batch_target_context",
        "assembler_alias": "ligand_library_loader",
        "integration_step": 0.01,
        "interpreter_alias": "ranking_interpreter"
      }
    }
  ],
  "wiring": [
    {
      "to": [
        "ligand_library_loader.scenario_context",
        "ranking_interpreter.scenario_context",
        "visualisation.batch_target_context_scenario_context"
      ],
      "from": "batch_target_context.scenario_context"
    },
    {
      "to": [
        "boltz_batch_ligand_ranker.protein_sequence"
      ],
      "from": "ligand_library_loader.protein_sequence"
    },
    {
      "to": [
        "boltz_batch_ligand_ranker.ligand_csv"
      ],
      "from": "ligand_library_loader.ligand_csv"
    },
    {
      "to": [
        "boltz_batch_ligand_ranker.msa_path"
      ],
      "from": "ligand_library_loader.msa_path"
    },
    {
      "to": [
        "boltz_batch_ligand_ranker.run_options"
      ],
      "from": "ligand_library_loader.run_options"
    },
    {
      "to": [
        "ranking_interpreter.assembled_boltz_request",
        "visualisation.ligand_library_loader_assembled_boltz_request"
      ],
      "from": "ligand_library_loader.assembled_boltz_request"
    },
    {
      "to": [
        "visualisation.boltz_batch_ligand_ranker_batch_summary",
        "ranking_interpreter.boltz_batch_ligand_ranker_batch_summary"
      ],
      "from": "boltz_batch_ligand_ranker.batch_summary"
    },
    {
      "to": [
        "visualisation.boltz_batch_ligand_ranker_affinity_summary",
        "ranking_interpreter.boltz_batch_ligand_ranker_affinity_summary"
      ],
      "from": "boltz_batch_ligand_ranker.affinity_summary"
    },
    {
      "to": [
        "visualisation.boltz_batch_ligand_ranker_confidence_summary",
        "ranking_interpreter.boltz_batch_ligand_ranker_confidence_summary"
      ],
      "from": "boltz_batch_ligand_ranker.confidence_summary"
    },
    {
      "to": [
        "visualisation.boltz_batch_ligand_ranker_structure_artifacts",
        "ranking_interpreter.boltz_batch_ligand_ranker_structure_artifacts"
      ],
      "from": "boltz_batch_ligand_ranker.structure_artifacts"
    },
    {
      "to": [
        "visualisation.boltz_batch_ligand_ranker_run_metadata",
        "ranking_interpreter.boltz_batch_ligand_ranker_run_metadata"
      ],
      "from": "boltz_batch_ligand_ranker.run_metadata"
    },
    {
      "to": [
        "visualisation.ranking_interpreter_prediction_evidence"
      ],
      "from": "ranking_interpreter.prediction_evidence"
    }
  ],
  "package": "boltz-batch-ligand-ranking-workflow",
  "runtime": {
    "duration": 0.01,
    "settle_steps": 1,
    "initial_inputs": {},
    "python_version": "3.10",
    "communication_step": 0.01
  },
  "version": "1.0.0",
  "description": "Guided Boltz-2 workflow for ranking a small ligand CSV against one protein target using binder probability, affinity-like value, confidence metrics, flags, and report-ready output.",
  "schema_version": "2.0"
}