skills.rest
openai
openaiOFFICIAL
Explore all skills by this author →
WebsiteGitHub

boltz-structure-and-binding

Predict protein, RNA, DNA, and ligand complex structures and binding metrics via the Boltz API.

5.3k|765|Updated Mar 4, 2026
Search Tags:#api#structure prediction#boltz#computational biology#binding affinity#protein folding#ligand docking
One-click install
npx skills add https://github.com/openai/plugins --skill boltz-structure-and-binding
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: boltz-structure-and-binding
Source: https://github.com/openai/plugins/tree/main/plugins/boltz-api-cli/skills/boltz-structure-and-binding
Command: npx skills add https://github.com/openai/plugins --skill boltz-structure-and-binding

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

Running a Boltz structure-and-binding prediction requires correctly authoring API payloads with exact field names, handling authentication, estimating cost, submitting jobs, and downloading results without blocking the agent session. This Skill encodes that entire workflow so a single complex prediction succeeds on the first attempt instead of failing on unclear 400 validation errors.

Core Features & Use Cases

  • Payload authoring: Normalizes proteins, RNA, DNA, and ligands (SMILES or CCD) into the entities schema with correct chain_ids arrays, optional constraints, bonds, modifications, templates, and MSA control.
  • Binding metrics: Adds flat binding blocks for ligand-protein or protein-protein binding, returning binding_confidence and optimization_score where applicable.
  • Managed job lifecycle: Runs estimate-cost for user confirmation, submits with an idempotency key, and launches download-results in the agent runtime's background mode with heartbeat-based status checks.
  • Use Case: A researcher wants to dock a small-molecule ligand against a target protein and get a binding confidence score. The Skill builds the payload, confirms cost, submits the job, and delivers the predicted CIF structure plus metrics.

Quick Start

Use the boltz-structure-and-binding skill to predict the structure of this protein-ligand complex and download the resulting CIF and metrics.

Frequently Asked Questions about boltz-structure-and-binding

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I predict a protein-ligand complex structure with Boltz?▼

Build a payload with protein and ligand_smiles entities using chain_ids arrays and value fields, then run boltz-api predictions:structure-and-binding estimate-cost, confirm the cost, and submit with start. Download results in the background with download-results using the same run name.

How do I get binding affinity metrics from a Boltz prediction?▼

Add a flat binding block to the payload with type ligand_protein_binding and a binder_chain_id pointing at the ligand chain. The results include binding_confidence and optimization_score under binding_metrics in metrics.json.

Why does the Boltz structure-and-binding API return a validation error with no details?▼

This endpoint can return only a generic VALIDATION_ERROR message without field-level paths. Inspect entities, binding, and constraints manually; common causes are using sequence instead of value, singular chain_id instead of chain_ids, or nesting the binding variant name.

Can Boltz binding predictions handle RNA or DNA complexes?▼

RNA and DNA entities are supported for structure prediction, but ligand-protein binding metrics only allow proteins and ligands in the entity set. Protein-protein binding uses binder_chain_ids and returns binding_confidence without an optimization_score.

What are the limitations of Boltz structure-and-binding predictions?▼

The skill is scoped to one defined complex, not library screening or molecular design. Ligand binders must have fewer than 50 atoms, atom-level bonds and contacts require ligand_ccd rather than ligand_smiles, and templates are limited to four CIF or PDB files.

Related Skills

Explore More Education & Research Skills→
InternScienceInternScience
28

molclaw-boltz2-affinity

Predict protein-ligand binding affinity from a protein sequence and SMILES using Boltz-2.

Official
Advanced
learningmatter-mitlearningmatter-mit
144

mat-xrd-phase-analysis

Identify crystalline phases in XRD patterns using DARA's Ray-based tree search.

Official
Advanced
InternScienceInternScience
28

molclaw-p2rank

Predicts ligand binding pockets in PDB proteins with confidence scores and docking boxes.

Official
Intermediate
adaptyvbioadaptyvbio
151

boltz

Predict protein complex structures with Boltz-1/Boltz-2 and output CIF models.

Official
Advanced