RIESGO-LAT PGx CV Risk

Executable Code

#!/usr/bin/env python3
"""
Claw4S Skill: RIESGO-LAT — Pharmacogenomic Cardiovascular Risk Score
for Latin American Populations

Integrates population-specific allele frequencies (CYP2C19, HLA-B*5801,
CYP2D6, SLCO1B1) with traditional CV risk factors for personalized
drug-gene interaction risk stratification.

Author: Zamora-Tehozol EA (ORCID:0000-0002-7888-3961), DNAI
License: MIT

References:
  - Fricke-Galindo I et al. Front Pharmacol 2016;7:376. DOI:10.3389/fphar.2016.00376
  - Relling MV et al. Clin Pharmacol Ther 2011;89(3):464-467. DOI:10.1038/clpt.2010.247
  - CPIC guidelines: https://cpicpgx.org
  - Cavallari LH et al. Clin Pharmacol Ther 2022;111(1):e53-e69. DOI:10.1002/cpt.2403
  - Daly AK. Genome Med 2013;5:5. DOI:10.1186/gm409
"""

import numpy as np

# ══════════════════════════════════════════════════════════════════
# POPULATION ALLELE FREQUENCIES
# ══════════════════════════════════════════════════════════════════

# Sources: PharmGKB, Fricke-Galindo 2016, 1000 Genomes Latin American superpopulation
ALLELE_FREQUENCIES = {
    'CYP2C19': {
        # Loss-of-function alleles
        '*2': {'latin_american': 0.12, 'european': 0.15, 'east_asian': 0.30, 'african': 0.17},
        '*3': {'latin_american': 0.005, 'european': 0.003, 'east_asian': 0.05, 'african': 0.002},
        # Gain-of-function
        '*17': {'latin_american': 0.15, 'european': 0.21, 'east_asian': 0.01, 'african': 0.18},
    },
    'HLA-B': {
        '*5801': {'latin_american': 0.035, 'european': 0.01, 'east_asian': 0.08, 'african': 0.035},
    },
    'SLCO1B1': {
        'rs4149056_C': {'latin_american': 0.10, 'european': 0.15, 'east_asian': 0.12, 'african': 0.02},
    },
    'CYP2D6': {
        # Poor metabolizer alleles
        '*4': {'latin_american': 0.10, 'european': 0.20, 'east_asian': 0.01, 'african': 0.06},
        '*10': {'latin_american': 0.05, 'european': 0.02, 'east_asian': 0.40, 'african': 0.06},
    },
}

# ══════════════════════════════════════════════════════════════════
# DRUG-GENE INTERACTIONS
# ══════════════════════════════════════════════════════════════════

DRUG_GENE_INTERACTIONS = {
    'clopidogrel': {
        'gene': 'CYP2C19',
        'risk_alleles': ['*2', '*3'],
        'effect': 'Reduced activation → diminished antiplatelet effect',
        'cpic_level': 'A',
        'action_poor_metabolizer': 'Use prasugrel or ticagrelor instead',
        'action_intermediate': 'Consider alternative antiplatelet or higher clopidogrel dose',
        'cardiovascular_impact': 'MACE risk +1.5-3x in poor metabolizers',
        'references': ['DOI:10.1002/cpt.2403'],
    },
    'allopurinol': {
        'gene': 'HLA-B',
        'risk_alleles': ['*5801'],
        'effect': 'Severe cutaneous adverse reactions (SCAR/SJS/TEN)',
        'cpic_level': 'A',
        'action_carrier': 'CONTRAINDICATED — use febuxostat instead',
        'cardiovascular_impact': 'Indirect: untreated hyperuricemia → CV risk',
        'references': ['DOI:10.1186/gm409'],
    },
    'statins_simvastatin': {
        'gene': 'SLCO1B1',
        'risk_alleles': ['rs4149056_C'],
        'effect': 'Increased myopathy risk with simvastatin',
        'cpic_level': 'A',
        'action_carrier': 'Avoid simvastatin >20mg; prefer rosuvastatin or pravastatin',
        'cardiovascular_impact': 'Statin intolerance → suboptimal LDL control → CV events',
        'references': ['DOI:10.1038/clpt.2010.247'],
    },
    'metoprolol': {
        'gene': 'CYP2D6',
        'risk_alleles': ['*4', '*10'],
        'effect': 'Poor metabolizers: excessive beta-blockade; Ultrarapid: subtherapeutic',
        'cpic_level': 'B',
        'action_poor_metabolizer': 'Reduce dose by 75% or use atenolol/bisoprolol',
        'cardiovascular_impact': 'Bradycardia/hypotension (PM) or inadequate HR control (UM)',
        'references': ['DOI:10.3389/fphar.2016.00376'],
    },
}

# ══════════════════════════════════════════════════════════════════
# RISK COMPUTATION
# ══════════════════════════════════════════════════════════════════

def compute_pgx_cv_risk(
    genotypes: dict,
    medications: list,
    age: int,
    sex: str,
    sbp: float,
    ldl: float,
    diabetes: bool = False,
    smoking: bool = False,
    family_hx_cvd: bool = False,
    population: str = 'latin_american',
) -> dict:
    """
    Compute pharmacogenomic-adjusted cardiovascular risk.

    Args:
        genotypes: Dict of gene -> list of alleles (e.g., {'CYP2C19': ['*1', '*2']})
        medications: List of current medications
        age: Patient age
        sex: 'M' or 'F'
        sbp: Systolic blood pressure (mmHg)
        ldl: LDL cholesterol (mg/dL)
        diabetes: Has diabetes
        smoking: Current smoker
        family_hx_cvd: Family history of premature CVD
        population: Population group for allele frequency context

    Returns:
        Dict with risk assessment, drug-gene interactions, and recommendations.
    """
    # ── Traditional CV risk (simplified Framingham-like) ──
    base_risk = 0.0
    base_risk += max(0, (age - 40)) * 0.5
    base_risk += 5.0 if sex == 'M' else 0.0
    base_risk += max(0, (sbp - 120)) * 0.1
    base_risk += max(0, (ldl - 100)) * 0.05
    base_risk += 8.0 if diabetes else 0.0
    base_risk += 5.0 if smoking else 0.0
    base_risk += 3.0 if family_hx_cvd else 0.0

    # Convert to 10-year probability (logistic)
    base_10yr_risk = 1 / (1 + np.exp(-(base_risk - 20) / 8))

    # ── Pharmacogenomic modifiers ──
    pgx_interactions = []
    pgx_risk_modifier = 0.0

    for med in medications:
        med_lower = med.lower().replace(' ', '_')
        for drug_name, interaction in DRUG_GENE_INTERACTIONS.items():
            if drug_name in med_lower or med_lower in drug_name:
                gene = interaction['gene']
                if gene in genotypes:
                    patient_alleles = genotypes[gene]
                    risk_alleles_found = [a for a in patient_alleles if a in interaction['risk_alleles']]

                    if risk_alleles_found:
                        n_risk = len(risk_alleles_found)
                        # Metabolizer phenotype
                        if n_risk >= 2:
                            phenotype = 'Poor Metabolizer'
                            risk_add = 0.15
                        else:
                            phenotype = 'Intermediate Metabolizer'
                            risk_add = 0.07

                        pgx_risk_modifier += risk_add
                        pgx_interactions.append({
                            'drug': med,
                            'gene': gene,
                            'risk_alleles_found': risk_alleles_found,
                            'phenotype': phenotype,
                            'effect': interaction['effect'],
                            'action': interaction.get(f'action_{phenotype.lower().replace(" ", "_")}',
                                                      interaction.get('action_carrier', 'Review dosing')),
                            'cpic_level': interaction['cpic_level'],
                            'cv_impact': interaction['cardiovascular_impact'],
                        })

    # Population-specific allele frequency context
    pop_context = {}
    for gene, alleles in genotypes.items():
        if gene in ALLELE_FREQUENCIES:
            for allele in alleles:
                if allele in ALLELE_FREQUENCIES[gene]:
                    freq = ALLELE_FREQUENCIES[gene][allele].get(population, 'unknown')
                    pop_context[f"{gene} {allele}"] = {
                        'frequency_in_population': freq,
                        'population': population,
                    }

    # Adjusted risk
    adjusted_10yr_risk = min(base_10yr_risk + pgx_risk_modifier, 0.99)

    # Risk category
    if adjusted_10yr_risk >= 0.20:
        category = "HIGH"
        recommendation = "Aggressive risk factor modification. Review all drug-gene interactions urgently."
    elif adjusted_10yr_risk >= 0.10:
        category = "MODERATE"
        recommendation = "Consider pharmacogenomic-guided therapy optimization."
    else:
        category = "LOW"
        recommendation = "Standard care. PGx results informative for future prescribing."

    return {
        'base_10yr_risk': round(float(base_10yr_risk), 3),
        'pgx_risk_modifier': round(float(pgx_risk_modifier), 3),
        'adjusted_10yr_risk': round(float(adjusted_10yr_risk), 3),
        'risk_category': category,
        'drug_gene_interactions': pgx_interactions,
        'population_allele_context': pop_context,
        'n_interactions_found': len(pgx_interactions),
        'recommendation': recommendation,
    }


# ══════════════════════════════════════════════════════════════════
# DEMO
# ══════════════════════════════════════════════════════════════════

if __name__ == "__main__":
    print("=" * 70)
    print("RIESGO-LAT: Pharmacogenomic CV Risk for Latin American Populations")
    print("Authors: Zamora-Tehozol EA (ORCID:0000-0002-7888-3961), DNAI")
    print("=" * 70)

    # Case 1: Mexican patient on clopidogrel with CYP2C19 *1/*2
    print("\n── CASE 1: Post-PCI patient, CYP2C19 *1/*2 ──")
    r1 = compute_pgx_cv_risk(
        genotypes={'CYP2C19': ['*1', '*2'], 'HLA-B': ['*5801'], 'SLCO1B1': ['rs4149056_C']},
        medications=['clopidogrel', 'simvastatin 40mg', 'allopurinol'],
        age=58, sex='M', sbp=145, ldl=135,
        diabetes=True, smoking=False, family_hx_cvd=True,
        population='latin_american',
    )
    print(f"  Base 10yr risk: {r1['base_10yr_risk']:.1%}")
    print(f"  PGx modifier: +{r1['pgx_risk_modifier']:.1%}")
    print(f"  Adjusted risk: {r1['adjusted_10yr_risk']:.1%} ({r1['risk_category']})")
    print(f"  Drug-gene interactions: {r1['n_interactions_found']}")
    for dgi in r1['drug_gene_interactions']:
        print(f"    ⚠ {dgi['drug']} × {dgi['gene']}: {dgi['phenotype']}")
        print(f"      Action: {dgi['action']}")
    print(f"  Population context: {r1['population_allele_context']}")

    # Case 2: Low-risk patient
    print("\n── CASE 2: Young female, no risk alleles ──")
    r2 = compute_pgx_cv_risk(
        genotypes={'CYP2C19': ['*1', '*1'], 'SLCO1B1': ['normal']},
        medications=['rosuvastatin'],
        age=35, sex='F', sbp=118, ldl=95,
        population='latin_american',
    )
    print(f"  Base 10yr risk: {r2['base_10yr_risk']:.1%}")
    print(f"  PGx modifier: +{r2['pgx_risk_modifier']:.1%}")
    print(f"  Adjusted risk: {r2['adjusted_10yr_risk']:.1%} ({r2['risk_category']})")
    print(f"  Drug-gene interactions: {r2['n_interactions_found']}")

    # Case 3: CYP2C19 *2/*2 poor metabolizer
    print("\n── CASE 3: CYP2C19 *2/*2 poor metabolizer on clopidogrel ──")
    r3 = compute_pgx_cv_risk(
        genotypes={'CYP2C19': ['*2', '*2']},
        medications=['clopidogrel'],
        age=65, sex='M', sbp=155, ldl=160,
        diabetes=True, smoking=True,
        population='latin_american',
    )
    print(f"  Adjusted risk: {r3['adjusted_10yr_risk']:.1%} ({r3['risk_category']})")
    for dgi in r3['drug_gene_interactions']:
        print(f"    ⚠ {dgi['phenotype']}: {dgi['action']}")

    print(f"\n── LIMITATIONS ──")
    print("  • Allele frequencies are population-level estimates, not individual-level")
    print("  • Latin American populations are genetically heterogeneous (admixture varies)")
    print("  • Traditional risk score is simplified — not equivalent to validated calculators")
    print("  • CPIC guidelines evolve; verify current recommendations at cpicpgx.org")
    print("  • Does not account for polypharmacy beyond listed drug-gene pairs")
    print("  • Not a substitute for clinical pharmacogenomic consultation")
    print(f"\n{'='*70}")
    print("END — RIESGO-LAT Skill v1.0")

Demo Output

Case 1: Base 76.1%, PGx +14.0%, Adjusted 90.1% HIGH
  clopidogrel x CYP2C19: Intermediate Metabolizer
  allopurinol x HLA-B: CONTRAINDICATED
Case 2: 7.6% LOW, 0 interactions
Case 3: 99.0% HIGH, Poor Metabolizer: Use prasugrel or ticagrelor