openbio_skill

Overview

This skill provides programmatic access to OpenBio services for biological data retrieval, literature search, structure prediction, pathway analysis, molecular biology design, variant and clinical data analysis, and plasmid file editing. It centralizes many bioinformatics tools behind a single API and guides you to validate parameters and handle long-running prediction jobs. You must set an OPENBIO_API_KEY and check the tool schema before invoking tools.

How this skill works

Authenticate with an API key stored in the OPENBIO_API_KEY environment variable. First query the tool schema for the specific tool to learn required parameter names, then validate parameters optionally before invoking the tool. Long-running prediction or submit_* tools return a job_id that you must poll for status and final results; many endpoints return download URLs for large outputs.

When to use it

• Query PDB, AlphaFold, UniProt, or other protein/structure resources
• Search literature across PubMed, bioRxiv, and arXiv
• Run structure prediction or sequence design (Boltz, Chai, ProteinMPNN, LigandMPNN)
• Perform pathway, network, or GO enrichment analysis
• Design molecular biology experiments: primers, cloning workflows, restriction digests, Gibson/Golden Gate
• Analyze variants, clinical data, or edit/parse plasmid files (GenBank, SnapGene)

Best practices

• Always GET the tool schema for the tool_name before calling it; parameter names often differ between tools
• Use the validate endpoint to pre-check parameters and avoid 400 errors
• For submit_* jobs, poll job status and only fetch results when status indicates completion
• Interpret returned quality metrics (pLDDT, resolution, GWAS p-values, Tanimoto) rather than using raw outputs blindly
• Store your OPENBIO_API_KEY securely in environment variables and create it at the OpenBio profile page if you don’t have one
• If API version is newer than your client, update the skill before continuing

Example use cases

• Fetch AlphaFold predictions and filter residues by pLDDT > 70 to identify confident regions
• Search PubMed for recent CRISPR reviews and download top abstracts for curation
• Submit a ProteinMPNN job to design sequences for a fixed backbone and poll until designed sequences are available
• Run GO enrichment on a differential gene list using g:Profiler endpoints and retrieve pathway networks via STRING
• Design PCR primers and simulate restriction enzyme digests for a planned cloning workflow
• Parse a SnapGene or GenBank plasmid file, modify features, and export an edited plasmid file