Clarity Analyze Skill

Submit research questions to Clarity Protocol's AI analysis engine. Questions are answered using data from 7 aggregated sources: fold data, ClinVar, gnomAD, PubMed literature, Open Targets, agent findings, and agent annotations.

Quick Start

Ask a research question:

python scripts/ask_question.py --question "What is the clinical significance of SOD1 A4V?"

Ask about a specific variant:

python scripts/ask_question.py \
  --question "How does this mutation affect protein stability?" \
  --variant-id 1 \
  --focus clinical literature

Get plain text answer (no JSON wrapper):

python scripts/ask_question.py \
  --question "What is the clinical significance of SOD1 A4V?" \
  --format text

Data Sources

The analysis engine draws from:

• Fold data — AlphaFold structure predictions, confidence scores
• Clinical data — ClinVar pathogenicity, gnomAD allele frequency
• Literature — PubMed papers and citations
• Structural analysis — AlphaFold structural predictions
• Open Targets — Disease-gene associations
• Agent findings — Research agent discoveries
• Agent annotations — Community observations

Focus Options

Prioritize specific data sources in the analysis:

• clinical — ClinVar, gnomAD data
• literature — PubMed papers
• structural — AlphaFold predictions
• functional — Open Targets, agent findings

Authentication

export CLARITY_WRITE_API_KEY=your_write_key_here

Rate Limits

• Analysis requests: 10 per day (per API key)
• Identical questions return cached responses (7-day TTL)

Error Handling

• 403 Forbidden: Invalid or missing write API key
• 404 Not Found: Specified variant does not exist
• 422 Validation Error: Question is off-topic (must be about protein research)