RIFT 2026 · Pharmacogenomics Intelligence v3
AI-driven genomic variant analysis using CPIC guidelines, DDGI phenoconversion engines, and Google Gemini 2.5 Flash — delivering clinical-grade drug safety insights from a single VCF file.
This dashboard details evidence-graded clinical recommendations for every drug-gene interaction identified in the patient's VCF. The engine evaluates the raw genetic diplotype and applies dynamic adjustments for Narrow Therapeutic Index (NTI) modifiers and Drug-Drug-Gene Interactions (DDGI) to calculate a precise 0-100% Risk Percentage.
| Gene / Variant | Drug | Effective Phenotype | Risk % | Recommendation |
|---|
Generative AI Summary: Medical jargon creates barriers to patient adherence.
Using the google-genai SDK, this module translates the deterministic CPIC outputs into
compassionate, plain-language education.
Pharmacogene Dictionary:
While genetic phenotypes are fixed, hepatic and renal clearance efficiency can be heavily modulated by lifestyle and diet. The AI engine generates personalized wellness protocols based on the specific enzymes flagged in the VCF.
| Modifier | Impacted Pathway |
|---|---|
| Grapefruit Juice | CYP3A4 / CYP1A2 Inhibition |
| Cruciferous Veggies | CYP1A2 Induction |
| St. John's Wort | CYP3A4 / CYP2C19 Broad Induction |
| High Vitamin K Greens | Warfarin Antagonism (VKORC1) |
| Alcohol Intake | Hepatic Glutathione Depletion |
The Phenoconversion Paradigm: A patient's genetically determined metabolizer status can be functionally overridden by a co-administered drug that strongly inhibits the same enzyme. For example, a genetic Normal Metabolizer taking Fluoxetine (a potent inhibitor) acts clinically as a Poor Metabolizer.
Deterministic Risk Percentage Engine:
| Phenotype Classification | Base Range | Clinical Rationale |
|---|---|---|
| Normal / Rapid Metabolizer | 5–15% | Expected PK; minimal background ADE risk |
| Intermediate Metabolizer | 35–55% | Variable exposure; dosage titration required |
| Poor / Ultrarapid Metabolizer | 82–99% | Severe PK disruption; predictable drug toxicity |
Machine Learning Subgrouping (Scikit-Learn):
KMeans(n_clusters=4, init='k-means++')StandardScaler().fit_transform(features)phenotype_score, risk_score, allele_frequencyNarrow Therapeutic Index (NTI) Weights:
Additive penalties are applied exclusively to non-normal phenotypes to heavily skew the risk score toward 100% for highly dangerous drugs.
Warfarin (+10) · 5-Fluorouracil (+8) · Capecitabine (+8) · Phenytoin (+7) · Azathioprine (+6) · Voriconazole (+5)
Zenomed is an investigational pharmacogenomic decision-support tool intended exclusively for use by licensed healthcare professionals and trained clinical pharmacologists. All outputs — including risk percentages, CPIC recommendations, AI explanations, and wellness tips — are AI-generated and must be independently verified before clinical application. This platform does not constitute medical advice, diagnosis, or treatment. Risk percentages are deterministic model outputs calibrated to population data and do not represent an individual patient's actual clinical outcome probability. Always consult a qualified physician, clinical pharmacologist, or certified genetic counselor before making therapeutic decisions based on pharmacogenomic data.