Genetics Interpretation Validation

Validate genetic variant interpretations by cross-referencing with SNPedia and clinical databases.

When to Use

• •Interpreting SNPs NOT in reference files
• •Verifying allele orientation (which genotype is risk allele)
• •Cross-checking genetic risk assessments
• •Investigating novel or rare variants
• •Confirming Gilbert syndrome, pharmacogenomics, or health risk interpretations

Workflow

1. •Receive SNP rsID and observed genotype
2. •Fetch SNPedia page for that SNP
3. •Extract genotype interpretations and clinical significance
4. •Compare observed genotype to reported risk/protective genotypes
5. •Return validated interpretation with confidence level

SNPedia Lookup

SNPedia URL format: https://www.snpedia.com/index.php/{rsid}

Example: https://www.snpedia.com/index.php/rs887829

Information to Extract

• •Genotype interpretations: (C;C), (C;T), (T;T) meanings
• •Risk allele: Which allele increases risk
• •Clinical significance: Magnitude and penetrance
• •Population frequency: How common is each genotype
• •Related conditions: Associated diseases/traits

Output Format

## SNP Validation: {rsid}

**Observed Genotype**: {genotype} (from 23andMe)

**SNPedia Interpretation**:
- **{genotype}**: {interpretation from SNPedia}
- **Risk Allele**: {allele}
- **Clinical Significance**: {description}

**Validation Status**: ✅ CONFIRMED / ⚠️ UNCERTAIN / ❌ CONTRADICTS

**Confidence**: HIGH / MODERATE / LOW

**Recommendation**:
{Any follow-up actions or caveats}

---

**Sources**:
- SNPedia: https://www.snpedia.com/index.php/{rsid}

Example Use Cases

Validate Gilbert Syndrome Interpretation

Input: rs887829 = CC Output:

✅ CONFIRMED: CC genotype is normal/wild-type
❌ INCORRECT prior interpretation as risk allele
Correct interpretation: TT = Gilbert syndrome risk

Check Unknown Pharmacogenomics SNP

Input: rs12345 = AG (not in reference files) Output:

⚠️ MODERATE CONFIDENCE: AG = intermediate metabolizer for CYP2C19
Recommend: Cross-check with clinical pharmacogenomics database

Integration Points

In investigate-root-cause skill

Add validation step after genetics analysis:

7. **Validate Genetics** (if genetics used): Use `genetics-validate-interpretation`
   skill to cross-check any SNP interpretations not in reference files

In genetics-snp-lookup skill

Add note to interpretation section:

⚠️ **If SNP not in reference files**: Use `genetics-validate-interpretation`
skill to fetch interpretation from SNPedia before making assumptions about allele orientation.

Error Handling

• •SNPedia page not found: Report "No SNPedia entry - interpretation uncertain"
• •Ambiguous results: Report "Multiple interpretations found - recommend clinical genetic testing"
• •Web fetch fails: Fallback to reference files only, note validation unavailable

Execution Instructions

1. •Receive input: rsID and genotype from user or calling skill
2. •Use WebFetch tool to fetch SNPedia page:
   code

   WebFetch(url="https://www.snpedia.com/index.php/{rsid}", prompt="Extract genotype interpretations, risk alleles, clinical significance, and population frequencies for this SNP")
   

3. •Parse response: Extract key information from SNPedia content
4. •Compare: Match observed genotype against SNPedia interpretations
5. •Format output: Use standard output format above
6. •Return: Validation status with confidence level

Key Validation Checks

Allele Orientation

• •Verify which allele is the risk allele (major vs minor)
• •Confirm genotype notation matches 23andMe format
• •Check for strand orientation issues (+ vs - strand)

Clinical Significance

• •Is this SNP clinically actionable?
• •What is the magnitude of effect?
• •What is the penetrance (how often does genotype cause phenotype)?

Population Context

• •How common is this genotype in the population?
• •Are there ethnic-specific interpretations?

Limitations

• •SNPedia coverage varies by SNP
• •Some SNPs have conflicting research findings
• •23andMe genotyping may miss important variants (e.g., UGT1A1*28)
• •Tag SNPs may not be the functional variant
• •Recommend clinical genetic testing for diagnostic confirmation

Advanced Validation: Allele Orientation Detection

The Strand Problem

23andMe reports genotypes on the forward (+) strand, but research literature may report on the reverse (-) strand. This causes confusion when matching genotypes to risk alleles.

Example: rs1801133 (MTHFR C677T)

• •23andMe reports: A/G (forward strand)
• •Some literature reports: T/C (reverse strand - complementary bases)
• •These are the same genotype on different strands

Complementary Base Pairs

Validation Workflow for Strand Issues

1. •

   Get genotype from 23andMe (always forward strand):

   bash

   grep "^rs1801133" "{genetics_23andme_path}"
   # Returns: rs1801133  1  11856378  AG
   

2. •

   Fetch SNPedia page and check reported alleles:

   code

   WebFetch(url="https://www.snpedia.com/index.php/rs1801133",
            prompt="What strand does this page report? List all genotype variations and their interpretations.")
   

3. •

   Compare and translate:

   markdown

   **23andMe (forward)**: A/G
   **SNPedia (if reverse)**: T/C (complementary)
   **Match**: A↔T, G↔C → Same genotype, different notation
   

4. •

   Report with clarification:

   markdown

   ## Strand Reconciliation: rs1801133
   
   | Source | Reported | Strand | Normalized |
   |--------|----------|--------|------------|
   | 23andMe | A/G | + (forward) | A/G |
   | SNPedia | T/C | - (reverse) | A/G |
   
   **Resolution**: Genotypes match after strand normalization.
   

Common SNPs with Strand Confusion

When to Suspect Strand Issues

• •Your genotype doesn't appear in SNPedia's genotype list
• •The reported risk allele is the complement of what you have
• •Literature uses gene-centric notation (e.g., "C677T") instead of rsID

Advanced Validation: Population Frequency Analysis

Why Population Frequency Matters

A rare genotype in European populations might be common in Asian populations. Interpretation must consider the patient's ancestry.

Workflow for Population Frequency Check

1. •

   Get genotype and patient ancestry from profile demographics

2. •

   Query SNPedia for frequency data:

   code

   WebFetch(url="https://www.snpedia.com/index.php/{rsid}",
            prompt="Extract population allele frequencies by ethnic group if available. Look for gnomAD, 1000 Genomes, or HapMap data.")
   

3. •

   Calculate genotype expected frequency:

   code

   For a genotype AA with allele frequency p(A):
   Expected frequency = p(A)² (Hardy-Weinberg)
   
   For heterozygote AG with p(A) and p(G):
   Expected frequency = 2 × p(A) × p(G)
   

4. •

   Interpret in context:

   markdown

   ## Population Context: rs429358 (APOE)
   
   **Your Genotype**: C/C (ε4/ε4)
   
   | Population | ε4 Frequency | ε4/ε4 Expected | Interpretation |
   |------------|--------------|----------------|----------------|
   | European | ~14% | ~2% | Uncommon but not rare |
   | African | ~20% | ~4% | More common |
   | East Asian | ~9% | ~0.8% | Relatively rare |
   | Finnish | ~23% | ~5% | Most common |
   
   **Patient ancestry**: European
   **Expected frequency**: ~2% (1 in 50)
   **Clinical note**: Uncommon genotype with highest Alzheimer's risk
   

Population Frequency Categories

Advanced Validation: Penetrance Assessment

What is Penetrance?

Penetrance = Probability that a person with a genotype will express the associated phenotype.

• •Complete penetrance (100%): Everyone with genotype has phenotype
• •High penetrance (>80%): Most people with genotype have phenotype
• •Reduced penetrance (20-80%): Variable expression
• •Low penetrance (<20%): Most carriers are unaffected

Penetrance Factors

1. •Age-dependent penetrance: Risk increases with age (e.g., BRCA, APOE)
2. •Sex-dependent penetrance: Different risk by gender (e.g., hemochromatosis)
3. •Modifier genes: Other variants affect expression
4. •Environment: Lifestyle, exposures, epigenetics

Penetrance Assessment Workflow

1. •

   Look up penetrance data:

   code

   WebFetch(url="https://www.snpedia.com/index.php/{rsid}",
            prompt="What is the penetrance of this variant? Is it age-dependent? Are there modifier factors?")
   

2. •

   Apply to patient context:

   markdown

   ## Penetrance Assessment: rs80357906 (BRCA1)
   
   **Variant**: BRCA1 185delAG (Ashkenazi founder)
   **Your Genotype**: Heterozygous carrier
   
   **Penetrance Data**:
   | Age | Breast Cancer Risk | Ovarian Cancer Risk |
   |-----|-------------------|---------------------|
   | 40 | 20% | 3% |
   | 50 | 40% | 10% |
   | 70 | 55-65% | 39-44% |
   
   **Modifiers**:
   - Family history of early-onset cancer → increases risk
   - BRCA2 co-carrier → substantially increases risk
   - Lifestyle factors (alcohol, obesity) → modulate risk
   
   **Clinical Note**: High but incomplete penetrance. Not all carriers develop cancer.
   Recommend: Genetic counseling, enhanced screening protocol.
   

3. •

   Quantify confidence based on penetrance:

   Penetrance	Risk Statement Confidence
   Complete (100%)	"Will develop" → HIGH
   High (>80%)	"Very likely to develop" → HIGH
   Moderate (50-80%)	"May develop" → MODERATE
   Reduced (20-50%)	"At increased risk" → MODERATE
   Low (<20%)	"Slightly increased risk" → LOW

Validation Checklist

Use this checklist for every genetic interpretation validation:

Pre-Validation Checks

• • Genotype confirmed: Verified genotype from 23andMe raw data
• • rsID format valid: Matches rs[0-9]+ pattern
• • SNPedia page exists: WebFetch returned content (not 404)

Strand and Allele Checks

• • Strand identified: Determined if SNPedia uses forward or reverse strand
• • Genotype translated: If strand mismatch, converted to matching notation
• • Risk allele identified: Know which allele increases/decreases risk
• • Genotype in SNPedia list: Your genotype appears in documented genotypes

Clinical Context Checks

• • Population frequency obtained: Have frequency for patient's ancestry
• • Penetrance assessed: Know likelihood genotype causes phenotype
• • Age-dependence noted: If applicable, risk by age documented
• • Modifier factors considered: Other genes, environment, lifestyle

Quality Checks

• • Multiple sources consulted: Not relying solely on SNPedia
• • Recent research checked: SNPedia may be outdated
• • Clinical actionability assessed: Is this finding clinically useful?
• • Limitations documented: What don't we know?

Output Requirements

• • Validation status clear: CONFIRMED / UNCERTAIN / CONTRADICTS
• • Confidence level stated: HIGH / MODERATE / LOW with reasoning
• • Sources cited: URLs to SNPedia pages and any other references
• • Recommendations included: What should user/provider do with this info?

Extended Error Handling

SNPedia Page Not Found

When SNPedia returns 404 or empty content:

## SNP Validation: rs999999999

**Status**: ⚠️ NOT FOUND IN SNPEDIA

**Possible reasons**:
1. **Invalid rsID**: Check for typos. Verify rsID in dbSNP.
2. **Very new SNP**: Recently discovered, not yet in SNPedia.
3. **Merged rsID**: Old rsID merged into another. Search dbSNP for current ID.
4. **Very rare variant**: Low research interest, limited documentation.

**Alternative lookup methods**:
1. **dbSNP**: https://www.ncbi.nlm.nih.gov/snp/{rsid}
2. **ClinVar**: https://www.ncbi.nlm.nih.gov/clinvar/?term={rsid}
3. **gnomAD**: https://gnomad.broadinstitute.org/variant/{rsid}
4. **PubMed search**: "rs999999999" or gene name + variant

**Recommendation**:
If SNP is clinically important, search alternative databases.
If still not found, interpretation confidence is VERY LOW.

Conflicting Information

When SNPedia shows contradictory research:

## SNP Validation: rs12345 - CONFLICTING DATA

**Your Genotype**: A/G

**Conflicting interpretations found**:

| Study | Year | Finding | Sample Size |
|-------|------|---------|-------------|
| Smith et al. | 2018 | AG = increased risk | n=500 |
| Jones et al. | 2021 | AG = no association | n=5000 |
| Lee et al. | 2023 | AG = protective | n=12000 |

**Resolution strategy**:
1. **Prefer larger studies**: Jones and Lee have more statistical power
2. **Prefer recent studies**: Lee 2023 is most current
3. **Check for confounders**: Different populations, methods, endpoints?
4. **Meta-analysis available?**: Look for systematic reviews

**Interpretation**: UNCERTAIN
More recent, larger studies suggest AG may be neutral or protective.
Earlier small study may have been false positive.

**Confidence**: LOW (conflicting evidence)

Malformed Cache Recovery

If cache file is corrupted:

cache_file=".claude/skills/health-agent/genetics-snp-lookup/.cache/${rsid}.json"

# Try to load cache
if ! python3 -c "import json; json.load(open('$cache_file'))" 2>/dev/null; then
  echo "WARNING: Corrupted cache for $rsid, removing and fetching fresh"
  rm -f "$cache_file"
  # Proceed to fresh fetch
fi

Network Timeout Handling

# Use timeout command (30 seconds default)
timeout 30 curl -A "health-agent/1.0" -s \
  "https://bots.snpedia.com/api.php?action=parse&page=${rsid}&format=json" \
  > "/tmp/claude/${rsid}_snpedia.json"

if [ $? -eq 124 ]; then
  echo "ERROR: SNPedia request timed out after 30s"
  # Try cache fallback
  if [ -f "$cache_file" ]; then
    echo "Using stale cache for $rsid"
    cat "$cache_file"
  else
    echo "No cache available, validation incomplete"
  fi
fi

Integration with Other Genetics Skills

Calling from investigate-root-cause

When the investigate-root-cause skill finds a potentially relevant SNP:

## Phase: Genetics Validation

For each SNP mentioned in hypothesis:

1. **Verify genotype** using `genetics-snp-lookup`
2. **Validate interpretation** using this skill (`genetics-validate-interpretation`)
3. **Document validation status** in hypothesis evidence

**Example integration**:
Hypothesis mentions rs887829 (Gilbert syndrome marker)
→ genetics-snp-lookup returns: CC genotype
→ genetics-validate-interpretation confirms:
   - SNPedia says CC = normal (NOT Gilbert)
   - Previous interpretation was WRONG
   - Adjusted hypothesis confidence accordingly

Calling from genetics-snp-lookup

When genetics-snp-lookup encounters ambiguous or unexpected results:

## Auto-Validation Triggers

genetics-snp-lookup should invoke genetics-validate-interpretation when:

1. **Genotype not in reference files**: Unknown clinical significance
2. **Strand confusion suspected**: Genotype doesn't match expected patterns
3. **User explicitly requests validation**: "Validate this interpretation"
4. **High-stakes interpretation**: BRCA, APOE, pharmacogenomics results

Calling from genetics-selfdecode-lookup

When comparing SelfDecode imputed data with SNPedia:

## Cross-Source Validation

1. **Get imputed genotype** from SelfDecode
2. **If 23andMe also has this SNP**, compare for concordance
3. **Validate interpretation** against SNPedia
4. **Note imputation confidence** in output

**Example**:
SelfDecode (imputed): rs12345 = AG
23andMe (direct): rs12345 = AG (if available)
SNPedia: AG interpretation
→ High confidence if all sources agree
→ Use 23andMe if sources disagree (direct genotyping > imputation)

Example: Complete Validation Workflow

User request: "Validate my interpretation of rs887829 for Gilbert syndrome"

Step 1: Get Genotype

grep "^rs887829" "{genetics_23andme_path}"
# Output: rs887829  2  234668879  CC

Step 2: Fetch SNPedia

WebFetch(url="https://www.snpedia.com/index.php/rs887829",
         prompt="Extract all genotype interpretations, risk alleles, and Gilbert syndrome association")

Step 3: Parse and Compare

SNPedia returns:

• •TT = Gilbert syndrome (elevated bilirubin)
• •CT = Carrier, mildly elevated bilirubin
• •CC = Normal (wild-type)

Patient genotype: CC

Step 4: Validate

## SNP Validation: rs887829

**Observed Genotype**: CC (from 23andMe)

**SNPedia Interpretation**:
- **TT**: Gilbert syndrome - Elevated unconjugated bilirubin
- **CT**: Carrier - Mildly elevated bilirubin
- **CC**: Normal - Wild-type UGT1A1 promoter

**Your Genotype (CC)**: Normal/wild-type. NOT associated with Gilbert syndrome.

**Validation Checklist**:
- [x] Genotype confirmed in raw data
- [x] SNPedia page found
- [x] Strand verified (forward strand, matches 23andMe)
- [x] Genotype in documented list
- [x] Clinical interpretation clear

**Validation Status**: ✅ CONFIRMED

**Confidence**: HIGH

**Clinical Note**:
If patient has elevated bilirubin, Gilbert syndrome via this SNP is ruled out.
Consider other causes: other UGT1A1 variants (not on 23andMe), hemolysis, liver disease.

**If prior interpretation said CC = Gilbert**: ❌ INCORRECT
The risk allele is T, not C. CC is wild-type (normal).

---

**Sources**:
- SNPedia: https://www.snpedia.com/index.php/rs887829
- 23andMe raw data: verified CC genotype