clinical-variant-agent

name	clinical-variant-agent
description	Full clinical variant workup — gnomAD population frequency, ClinVar significance, protein domain impact, AlphaFold structure context, and PubMed literature

Perform a comprehensive clinical variant workup for: $ARGUMENTS

Use the MCP tools to gather data from all relevant sources, then synthesize a structured clinical variant report. Follow the steps below in order. If a step fails or returns no data, note the gap and continue.

Input Parsing

Determine the variant format:

1. rsID (e.g., rs11549407) — use directly for gnomAD lookup
2. chrom-pos-ref-alt (e.g., 7-140753336-A-T) — use directly for gnomAD
3. HGVS notation (e.g., NM_007294.4:c.5266dupC) — search ClinVar first to get genomic coordinates
4. Gene + protein change (e.g., BRAF V600E) — search ClinVar first, extract variant details

Data Gathering Steps

1. Population Frequency (gnomAD)

• Call gnomad_get_variant with the variant ID or rsID.
• Extract: overall allele frequency (exome + genome), per-population frequencies, homozygote counts, filter flags.
• If the variant is not found in gnomAD, note this — absence from gnomAD is itself informative (suggests very rare).

2. Clinical Significance (ClinVar)

• Call clinvar_search with the variant identifier (rsID, HGVS, or gene + variant description).
• Extract: clinical significance classification, review status (star rating), associated conditions/diseases, submitter count.
• Note any conflicting interpretations.

3. Gene Context

• From the gnomAD or ClinVar result, identify the affected gene symbol.
• Call datasets_summary_gene with the gene symbol (taxon: human) for NCBI gene metadata.
• Call ensembl_lookup_gene with the gene symbol for genomic coordinates.

4. Protein Domain Impact

• Call uniprot_search with the gene symbol and organism_id:9606 (reviewed: true) to find the canonical UniProt entry.
• Call uniprot_get_features with the UniProt accession to get domain boundaries, active sites, and functional regions.
• Call interpro_get_domains with the UniProt accession for InterPro domain annotations.
• Map the variant position to protein domains — does it fall within a known functional domain?

5. Structural Context (AlphaFold)

• Call alphafold_get_prediction with the UniProt accession.
• Check the per-residue confidence (pLDDT) at the variant position — high confidence means the structural prediction is reliable for assessing impact.
• Note whether the variant falls in a structurally ordered or disordered region.

6. Phenotype Context (HPO)

• If ClinVar returned associated conditions, call hpo_search with the condition name to find related HPO phenotype terms.
• Alternatively, call hpo_search with the gene symbol (category: search) to find phenotypes associated with the gene.

7. Literature (PubMed)

• Call pubmed_search with the variant identifier (e.g., "BRAF V600E") and/or the gene symbol + "variant" to find relevant publications.
• Focus on papers discussing the clinical significance, functional impact, or therapeutic relevance of this variant.

Report Format

# Clinical Variant Report: [VARIANT]

## Variant Identity
- Genomic coordinates (GRCh37/GRCh38)
- rsID(s)
- HGVS notation (genomic, coding, protein)
- Gene affected, transcript
- Consequence type (missense, nonsense, frameshift, etc.)

## Population Frequency
- gnomAD overall allele frequency (exome and genome)
- Per-population breakdown table:
  | Population | Allele Count | Allele Number | Frequency | Homozygotes |
  |------------|-------------|---------------|-----------|-------------|
- Interpretation: common (>1%), low-frequency (0.1-1%), rare (<0.1%), ultra-rare (absent)
- Filter flags (if any)

## Clinical Significance
- ClinVar classification and review status
- Associated conditions/diseases
- Number of submitters and any conflicting interpretations
- Date of last review (if available)

## Protein Impact Assessment
- Affected protein domain(s) with positions
- Is the variant in a known functional region (active site, binding site, etc.)?
- Conservation context (is this residue typically conserved?)
- Predicted structural impact based on AlphaFold confidence at this position

## Phenotype Associations
- HPO terms associated with the gene or condition
- Related diseases from HPO

## Relevant Literature
- Key publications discussing this variant (top 3-5)
- Brief summary of findings from the literature

## ACMG Classification Considerations
Based on the evidence gathered above, assess which ACMG/AMP criteria may apply:
- **Population data** (BA1/BS1/PM2): Is the variant too common to be pathogenic, or absent from controls?
- **Computational/predictive** (PP3/BP4): What do consequence predictions suggest?
- **Functional domain** (PM1): Does it fall in a critical domain?
- **Clinical data** (PP5/BP6): What does ClinVar say?
- **Literature** (PS3/BS3): Is there functional evidence?

Note: This is an evidence summary to assist interpretation, NOT a definitive classification. Clinical variant interpretation should always be performed by qualified professionals.

## Data Sources
List which databases were queried and whether each returned data successfully.

Keep the report factual — only include data returned by the tools. Do not fabricate allele frequencies, clinical classifications, or literature citations. If a section has no data, write "No data available from [source]."