tooluniverse-variant-interpretation

---

name: tooluniverse-variant-interpretation description: Systematic clinical variant interpretation from raw variant calls to ACMG-classified recommendations with structural impact analysis. Aggregates evidence from ClinVar, gnomAD, CIViC, UniProt, and PDB across ACMG criteria. Produces pathogenicity scores (0-100), clinical recommendations, and treatment implications. Use when interpreting genetic variants, classifying variants of uncertain significance (VUS), performing ACMG variant classification, or translating variant calls to clinical actionability.

Clinical Variant Interpreter

Systematic variant interpretation skill using ToolUniverse - from raw variant calls to ACMG-classified clinical recommendations with structural impact analysis.

---

Problem This Skill Solves

Clinical labs and researchers face critical challenges in variant interpretation:

1. Variant classification uncertainty - VUS (Variants of Uncertain Significance) comprise 40-60% of clinical variants
2. Evidence aggregation burden - Must integrate data from 10+ databases per variant
3. Structural context missing - Traditional annotation ignores 3D protein impact
4. Clinical actionability unclear - How does classification translate to patient care?

This skill provides: A systematic workflow that combines population databases, functional predictions, structural analysis (via AlphaFold2), and literature evidence into ACMG-compliant interpretations with clear clinical recommendations.

---

Key Principles

1. ACMG-Guided Classification - Follow ACMG/AMP 2015 guidelines with explicit evidence codes
2. Structural Evidence Integration - Use AlphaFold2 for novel structural impact analysis
3. Population Context - gnomAD frequencies with ancestry-specific data
4. Gene-Disease Validity - ClinGen curation status for clinical relevance
5. Actionable Output - Clear recommendations, not just classifications
6. English-first queries - Always use English terms in tool calls (gene names, variant descriptions, disease names), even if the user writes in another language. Only try original-language terms as a fallback. Respond in the user's language

---

Triggers

Use this skill when users:

• Ask about variant interpretation or classification
• Have VCF data needing clinical annotation
• Ask "what does this variant mean clinically?"
• Need ACMG classification for variants
• Want structural impact analysis for missense variants
• Ask about pathogenicity of specific variants

---

Workflow Overview

┌─────────────────────────────────────────────────────────────────┐
│                    VARIANT INTERPRETATION                        │
├─────────────────────────────────────────────────────────────────┤
│                                                                  │
│  Phase 1: VARIANT IDENTITY                                       │
│  ├── Normalize variant notation (HGVS)                          │
│  ├── Map to gene, transcript, protein                           │
│  └── Get consequence type (missense, nonsense, etc.)            │
│                                                                  │
│  Phase 2: CLINICAL DATABASES                                     │
│  ├── ClinVar: Existing classifications                          │
│  ├── gnomAD: Population frequencies (all + ancestry)            │
│  ├── OMIM: Gene-disease associations                            │
│  ├── ClinGen: Gene validity + dosage sensitivity (ENHANCED)     │
│  │   └─ ClinGen_search_gene_validity, ClinGen_search_dosage     │
│  └── SpliceAI: Splice variant prediction (NEW)                  │
│                                                                  │
│  Phase 2.5: REGULATORY CONTEXT (NEW - for non-coding variants)  │
│  ├── ChIPAtlas: TF binding at position                          │
│  ├── ENCODE: Regulatory elements (enhancers, promoters)         │
│  ├── Conservation in regulatory regions                         │
│  └── Functional annotation of regulatory impact                 │
│                                                                  │
│  Phase 3: COMPUTATIONAL PREDICTIONS                              │
│  ├── SIFT/PolyPhen: Damaging predictions                        │
│  ├── CADD: Deleteriousness score                                │
│  ├── SpliceAI: Splice impact (if applicable)                    │
│  └── Conservation: Cross-species alignment                      │
│                                                                  │
│  Phase 4: STRUCTURAL ANALYSIS (for VUS/novel missense)          │
│  ├── Get protein structure (PDB or AlphaFold2)                  │
│  ├── Map variant to structure                                   │
│  ├── Assess domain/functional site impact                       │
│  └── Predict structural destabilization                         │
│                                                                  │
│  Phase 4.5: EXPRESSION CONTEXT (NEW)                            │
│  ├── CELLxGENE: Cell-type specific expression                   │
│  ├── Tissue relevance to phenotype                              │
│  └── Expression validation                                       │
│                                                                  │
│  Phase 5: LITERATURE EVIDENCE                                    │
│  ├── PubMed: Functional studies                                 │
│  ├── BioRxiv/MedRxiv: Recent preprints (NEW)                   │
│  ├── Case reports: Phenotype correlations                       │
│  └── Segregation data (if in literature)                        │
│                                                                  │
│  Phase 6: ACMG CLASSIFICATION                                    │
│  ├── Apply evidence codes (PVS1, PM2, PP3, etc.)               │
│  ├── Calculate classification                                   │
│  ├── Identify limiting factors                                  │
│  └── Generate clinical recommendations                          │
│                                                                  │
└─────────────────────────────────────────────────────────────────┘

---

Phase Details

Phase 1: Variant Identity & Normalization

Goal: Standardize variant notation and determine molecular consequence

Tools:

Tool	Purpose
myvariant_query	Get variant annotations from MyVariant.info
Ensembl_get_variant_info	Variant effect predictor data
NCBI_gene_search	Gene information

Key Information to Capture:

• HGVS notation (c. and p.)
• Gene symbol and Ensembl ID
• Transcript (canonical/MANE Select)
• Consequence type
• Amino acid change (for missense)
• Exon/intron location

Phase 2: Clinical Database Queries

Goal: Aggregate existing clinical knowledge

Tools:

Tool	Purpose	Key Data
clinvar_search	Existing classifications	Classification, review status, submissions
gnomad_search	Population frequency	AF, ancestry-specific AFs, homozygotes
OMIM_search, OMIM_get_entry	Gene-disease	Inheritance, phenotypes
ClinGen_gene_validity	Curation status	Gene-disease validity level
COSMIC_search_mutations	Somatic mutations (NEW)	Cancer frequency, histology
DisGeNET_search_gene	Gene-disease associations (NEW)	Evidence scores, sources

2.1 COSMIC for Somatic Context (NEW)

For cancer variants, check COSMIC for somatic mutation frequency:

def get_somatic_context(tu, gene_symbol, variant_aa):
    """Get somatic mutation context from COSMIC."""

    # Search for specific mutation
    cosmic = tu.tools.COSMIC_search_mutations(
        operation="search",
        terms=f"{gene_symbol} {variant_aa}",
        max_results=20,
        genome_build=38
    )

    # Get all gene mutations for context
    gene_mutations = tu.tools.COSMIC_get_mutations_by_gene(
        operation="get_by_gene",
        gene=gene_symbol,
        max_results=100
    )

    # Determine if it's a hotspot
    mutation_counts = Counter(m['MutationAA'] for m in gene_mutations.get('results', []))
    is_hotspot = variant_aa in [m[0] for m in mutation_counts.most_common(10)]

    return {
        'cosmic_hits': cosmic.get('results', []),
        'is_somatic_hotspot': is_hotspot,
        'cancer_types': [m['PrimarySite'] for m in cosmic.get('results', [])],
        'total_cosmic_count': cosmic.get('total_count', 0)
    }

2.2 OMIM Gene-Disease Context (NEW)

def get_omim_context(tu, gene_symbol):
    """Get OMIM gene-disease associations."""

    # Search OMIM for gene
    search = tu.tools.OMIM_search(
        operation="search",
        query=gene_symbol,
        limit=5
    )

    omim_data = []
    for entry in search.get('data', {}).get('entries', []):
        mim = entry.get('mimNumber')

        # Get detailed entry
        details = tu.tools.OMIM_get_entry(
            operation="get_entry",
            mim_number=str(mim)
        )

        # Get clinical synopsis
        synopsis = tu.tools.OMIM_get_clinical_synopsis(
            operation="get_clinical_synopsis",
            mim_number=str(mim)
        )

        omim_data.append({
            'mim_number': mim,
            'title': details.get('data', {}).get('titles', {}),
            'inheritance': synopsis.get('data', {}).get('inheritance'),
            'clinical_features': synopsis.get('data', {})
        })

    return omim_data

2.3 DisGeNET Gene-Disease Evidence (NEW)

def get_disgenet_context(tu, gene_symbol, variant_rsid=None):
    """Get gene-disease associations from DisGeNET."""

    # Gene-disease associations
    gda = tu.tools.DisGeNET_search_gene(
        operation="search_gene",
        gene=gene_symbol,
        limit=20
    )

    # Variant-disease associations (if rsID available)
    vda = None
    if variant_rsid:
        vda = tu.tools.DisGeNET_get_vda(
            operation="get_vda",
            variant=variant_rsid,
            limit=20
        )

    return {
        'gene_associations': gda.get('data', {}).get('associations', []),
        'variant_associations': vda.get('data', {}).get('associations', []) if vda else []
    }

2.4 ClinGen Gene Validity & Dosage Sensitivity (NEW)

ClinGen provides authoritative curation of gene-disease relationships:

def get_clingen_evidence(tu, gene_symbol):
    """
    Get ClinGen gene validity and dosage sensitivity data.
    CRITICAL for ACMG classification - establishes gene-disease validity.
    """

    # 1. Gene-disease validity (Definitive/Strong/Moderate/Limited)
    validity = tu.tools.ClinGen_search_gene_validity(gene=gene_symbol)

    validity_data = []
    if validity.get('data'):
        for entry in validity.get('data', []):
            validity_data.append({
                'disease': entry.get('Disease Label'),
                'classification': entry.get('Classification'),  # Definitive, Strong, etc.
                'inheritance': entry.get('Inheritance'),
                'mondo_id': entry.get('Disease ID (MONDO)')
            })

    # 2. Dosage sensitivity (haploinsufficiency, triplosensitivity)
    dosage = tu.tools.ClinGen_search_dosage_sensitivity(gene=gene_symbol)

    dosage_data = {}
    if dosage.get('data'):
        for entry in dosage.get('data', []):
            dosage_data = {
                'haploinsufficiency_score': entry.get('Haploinsufficiency Score'),
                'triplosensitivity_score': entry.get('Triplosensitivity Score'),
                'disease': entry.get('Disease')
            }
            break  # Usually one entry per gene

    # 3. Clinical actionability (for incidental findings context)
    actionability = tu.tools.ClinGen_search_actionability(gene=gene_symbol)

    return {
        'gene_validity': validity_data,
        'dosage_sensitivity': dosage_data,
        'actionability': actionability.get('data', {}),
        'has_definitive_validity': any(v['classification'] == 'Definitive' for v in validity_data),
        'is_haploinsufficient': dosage_data.get('haploinsufficiency_score') == '3'
    }

ClinGen Validity Levels (for ACMG PM1/PP4):

Classification	Meaning	ACMG Impact
Definitive	Multiple concordant studies	Strong gene-disease support
Strong	Extensive evidence	Moderate-strong support
Moderate	Some evidence	Moderate support
Limited	Minimal evidence	Weak support, use caution
Disputed	Conflicting evidence	Do not use for classification
Refuted	Evidence against	Gene NOT associated

Dosage Sensitivity Scores (for CNV interpretation):

Score	Meaning	Interpretation
3	Sufficient evidence	Haploinsufficiency/triplosensitivity established
2	Emerging evidence	Some support, not definitive
1	Little evidence	Minimal support
0	No evidence	Unknown

2.5 SpliceAI Splice Variant Prediction (NEW)

~15% of pathogenic variants affect splicing. SpliceAI is the gold standard for splice prediction:

def get_spliceai_prediction(tu, chrom, pos, ref, alt, genome="38"):
    """
    Get SpliceAI splice effect predictions.

    Delta scores:
    - DS_AG: Acceptor gain
    - DS_AL: Acceptor loss
    - DS_DG: Donor gain
    - DS_DL: Donor loss

    Thresholds:
    - ≥0.8: High pathogenicity (strong PP3)
    - 0.5-0.8: Moderate (supporting PP3)
    - 0.2-0.5: Low (weak evidence)
    - <0.2: Likely benign
    """

    # Format variant for SpliceAI
    variant = f"chr{chrom}-{pos}-{ref}-{alt}"

    # Get full splice predictions
    result = tu.tools.SpliceAI_predict_splice(
        variant=variant,
        genome=genome
    )

    if result.get('data'):
        max_score = result['data'].get('max_delta_score', 0)
        interpretation = result['data'].get('interpretation', '')

        # Determine ACMG support
        if max_score >= 0.8:
            acmg = 'PP3 (strong) - high splice impact'
        elif max_score >= 0.5:
            acmg = 'PP3 (supporting) - moderate splice impact'
        elif max_score >= 0.2:
            acmg = 'PP3 (weak) - possible splice impact'
        else:
            acmg = 'BP7 (if synonymous) - splice benign'

        return {
            'max_delta_score': max_score,
            'interpretation': interpretation,
            'acmg_support': acmg,
            'scores': result['data'].get('scores', [])
        }
    return None

def quick_splice_check(tu, variant, genome="38"):
    """Quick triage using max delta score only."""

    result = tu.tools.SpliceAI_get_max_delta(
        variant=variant,
        genome=genome
    )

    return result.get('data', {})

When to Use SpliceAI:

• Intronic variants near splice sites (±50bp)
• Synonymous variants (may still affect splicing)
• Exonic variants near splice junctions
• Variants creating cryptic splice sites

Report Section for Splice Variants:

### Splice Impact Analysis (SpliceAI)

| Score Type | Value | Position | Interpretation |
|------------|-------|----------|----------------|
| DS_AG | 0.02 | +15 | Acceptor gain unlikely |
| DS_AL | 0.85 | -2 | **High acceptor loss** |
| DS_DG | 0.01 | +8 | Donor gain unlikely |
| DS_DL | 0.03 | +1 | Donor loss unlikely |

**Max Delta Score**: 0.85 (DS_AL)
**Interpretation**: High impact - likely disrupts acceptor site
**ACMG Support**: PP3 (strong) for splice-altering effect

*Source: SpliceAI via `SpliceAI_predict_splice`*

ClinVar Classification Map:

ClinVar	Interpretation
Pathogenic	Disease-causing
Likely pathogenic	90%+ confidence pathogenic
VUS	Uncertain significance
Likely benign	90%+ confidence benign
Benign	Not disease-causing
Conflicting	Multiple interpretations

gnomAD Thresholds (for rare disease):

Frequency	ACMG Code	Interpretation
Absent	PM2_Supporting	Absent from controls
<0.00001	PM2_Supporting	Extremely rare
<0.0001	-	Rare (use with caution)
>0.01	BS1/BA1	Too common for rare disease

COSMIC Somatic Evidence (NEW):

COSMIC Finding	Interpretation	ACMG Support
Recurrent hotspot (>100 samples)	Known oncogenic driver	PS3 (functional)
Moderate frequency (10-100)	Likely oncogenic	PM1 (hotspot)
Rare somatic (<10)	Unknown significance	No support

DisGeNET Score Interpretation (NEW):

GDA Score	Evidence Level	ACMG Support
>0.7	Strong	PP4 (phenotype)
0.4-0.7	Moderate	Supporting
<0.4	Weak	Insufficient

Phase 2.5: Regulatory Context (NEW - for Non-Coding Variants)

Goal: Assess regulatory impact for non-coding, intronic, and promoter variants

When to Apply:

• Intronic variants (not splice site)
• Promoter variants
• 5'UTR / 3'UTR variants
• Intergenic variants near disease genes

Tools:

Tool	Purpose	Key Data
ChIPAtlas_enrichment_analysis	TF binding at position	Bound TFs, cell types
ChIPAtlas_get_peak_data	ChIP-seq peaks	Peak coordinates, scores
ENCODE_search_experiments	Regulatory elements	Enhancers, promoters, DHS
ENCODE_get_experiment	Experiment details	Assay type, targets

Regulatory Impact Assessment:

def assess_regulatory_impact(tu, variant_position, gene_symbol):
    """Assess regulatory impact of non-coding variant."""

    # Check TF binding at position
    tf_binding = tu.tools.ChIPAtlas_enrichment_analysis(
        gene=gene_symbol,
        cell_type="all"
    )

    # Get ChIP-seq peaks overlapping variant
    peaks = tu.tools.ChIPAtlas_get_peak_data(
        gene=gene_symbol,
        experiment_type="TF"
    )

    # Search ENCODE for regulatory annotations
    encode_data = tu.tools.ENCODE_search_experiments(
        assay_title="ATAC-seq",
        biosample="all"
    )

    # Assess if variant disrupts TF binding
    binding_disrupted = check_motif_disruption(variant_position, peaks)

    return {
        'tf_binding': tf_binding,
        'regulatory_peaks': peaks,
        'encode_annotations': encode_data,
        'likely_regulatory': binding_disrupted
    }

Regulatory Impact Categories:

Category	Criteria	ACMG Support
High impact	Disrupts known TF binding motif	PP3 (supporting)
Moderate impact	In active regulatory region	Consider context
Low impact	No regulatory annotation	No support

Output for Report:

### 2.5 Regulatory Context (for Non-Coding Variants)

| Feature | Finding | Significance |
|---------|---------|--------------|
| Variant location | Intron 5, 120bp from exon 6 | Not canonical splice |
| TF binding site | CTCF binding peak (ChIPAtlas) | May affect insulation |
| ENCODE annotation | Active enhancer (H3K27ac) | Regulatory function |
| Conservation | PhyloP = 2.8 | Moderate conservation |

**Regulatory Interpretation**: Variant overlaps CTCF binding site in active enhancer region. Potential impact on gene regulation.

*Source: ChIPAtlas, ENCODE*

Phase 3: Computational Predictions (ENHANCED)

Goal: Assess in silico pathogenicity predictions using state-of-the-art models

Tools:

Tool	Purpose	Score Range
CADD_get_variant_score	Deleteriousness score (NEW API)	PHRED 0-99
AlphaMissense_get_variant_score	DeepMind pathogenicity (NEW)	0-1
EVE_get_variant_score	Evolutionary pathogenicity (NEW)	0-1
myvariant_query	Aggregated predictions	SIFT, PolyPhen
Ensembl_get_variant_info	VEP predictions	SIFT, PolyPhen

3.1 CADD Deleteriousness Scoring (NEW)

def get_cadd_score(tu, chrom, pos, ref, alt):
    """Get CADD deleteriousness score for a variant."""

    result = tu.tools.CADD_get_variant_score(
        chrom=str(chrom),
        pos=pos,
        ref=ref,
        alt=alt,
        version="GRCh38-v1.7"
    )

    if result.get('status') == 'success':
        phred = result['data'].get('phred_score')
        return {
            'score': phred,
            'interpretation': result['data'].get('interpretation'),
            'acmg_support': 'PP3' if phred >= 20 else ('BP4' if phred < 15 else 'neutral')
        }
    return None

3.2 AlphaMissense Pathogenicity (NEW)

DeepMind's AlphaMissense provides state-of-the-art missense pathogenicity prediction:

def get_alphamissense_score(tu, uniprot_id, variant):
    """
    Get AlphaMissense pathogenicity score.
    variant format: 'R123H' or 'p.R123H'

    Thresholds:
    - Pathogenic: score > 0.564
    - Ambiguous: 0.34-0.564
    - Benign: score < 0.34
    """

    result = tu.tools.AlphaMissense_get_variant_score(
        uniprot_id=uniprot_id,
        variant=variant
    )

    if result.get('status') == 'success' and result.get('data'):
        score = result['data'].get('pathogenicity_score')
        classification = result['data'].get('classification')

        # Map to ACMG
        if classification == 'pathogenic':
            acmg = 'PP3 (strong)'  # AlphaMissense has high accuracy
        elif classification == 'benign':
            acmg = 'BP4 (strong)'
        else:
            acmg = 'neutral'

        return {
            'score': score,
            'classification': classification,
            'acmg_support': acmg
        }
    return None

3.3 EVE Evolutionary Prediction (NEW)

EVE uses unsupervised learning on evolutionary data:

def get_eve_score(tu, chrom, pos, ref, alt):
    """
    Get EVE evolutionary pathogenicity score.

    Threshold: >0.5 indicates likely pathogenic
    """

    result = tu.tools.EVE_get_variant_score(
        chrom=str(chrom),
        pos=pos,
        ref=ref,
        alt=alt
    )

    if result.get('status') == 'success':
        eve_scores = result['data'].get('eve_scores', [])
        if eve_scores:
            best_score = eve_scores[0]
            return {
                'score': best_score.get('eve_score'),
                'classification': best_score.get('classification'),
                'gene': best_score.get('gene_symbol'),
                'acmg_support': 'PP3' if best_score.get('eve_score', 0) > 0.5 else 'BP4'
            }
    return None

3.4 Integrated Prediction Strategy

For VUS (Variants of Uncertain Significance), combine multiple predictors:

def comprehensive_pathogenicity_assessment(tu, variant_info):
    """
    Combine all prediction tools for robust classification.
    """
    chrom = variant_info['chrom']
    pos = variant_info['pos']
    ref = variant_info['ref']
    alt = variant_info['alt']
    uniprot_id = variant_info.get('uniprot_id')
    aa_change = variant_info.get('aa_change')  # e.g., 'R123H'

    predictions = {}

    # 1. CADD (works for all variant types)
    cadd = get_cadd_score(tu, chrom, pos, ref, alt)
    if cadd:
        predictions['cadd'] = cadd

    # 2. AlphaMissense (missense only, requires UniProt ID)
    if uniprot_id and aa_change:
        am = get_alphamissense_score(tu, uniprot_id, aa_change)
        if am:
            predictions['alphamissense'] = am

    # 3. EVE (missense only)
    eve = get_eve_score(tu, chrom, pos, ref, alt)
    if eve:
        predictions['eve'] = eve

    # Consensus assessment
    damaging_count = sum(1 for p in predictions.values()
                         if 'PP3' in p.get('acmg_support', ''))
    benign_count = sum(1 for p in predictions.values()
                       if 'BP4' in p.get('acmg_support', ''))