PharmaGuard 🧬💊

CPIC-Aligned Pharmacogenomic Risk Prediction Platform

🔗 Quick Links

Resource	Link
🌐 Live Demo	[https://pharma-guard-topaz.vercel.app/]
🎥 Demo Video (LinkedIn)	[https://www.linkedin.com/posts/fascinate_rift2026-rifthackathon-pharmacogenomics-activity-7430419912659218432-zypO?utm_source=share&utm_medium=member_desktop&rcm=ACoAAEesNGIBK6EVCjHwL6vDr5aFMZkoVPuhDMQ]
📂 GitHub Repository	[https://github.com/fscntmohit/Pharma-Guard.git]

🎯 Overview

PharmaGuard is a precision medicine web application that analyzes patient genetic data (VCF files) to predict drug-gene interaction risks using CPIC-aligned rule-based logic. The system provides clinically actionable recommendations with AI-powered explanations.

Key Differentiators

Rule-Based Phenotype Classification: Deterministic CPIC-aligned diplotype → phenotype mapping (no LLM guessing)
CPIC-Aligned Risk Engine: Drug-specific risk rules following clinical guidelines
LLM for Explanation Only: AI generates explanations but never makes clinical decisions
Consistent Clinical Output: Phenotype → Risk → Recommendation are always internally consistent

✨ Features

🧬 VCF File Upload - Drag & drop or file picker with validation
💊 6 Critical Drug-Gene Pairs - CPIC Level A evidence drugs
🎯 CPIC-Aligned Phenotyping - Strict rule-based diplotype interpretation
⚠️ Risk Stratification - Safe, Adjust Dosage, Toxic, Ineffective classifications
🤖 AI Explanations - GPT-powered clinical context (explanation only, not decision)
📊 JSON Export - Competition-compliant schema output
📱 Responsive Design - Modern healthcare UI with Tailwind CSS

🛠 Tech Stack

Layer	Technology
Frontend	React 18 + Tailwind CSS 3.4
Backend	Node.js + Express
AI/LLM	OpenAI API (GPT-3.5-turbo)
Parsing	Custom VCF Parser
Logic	CPIC-Aligned Rule Engines

📁 Project Structure

PharmaGuard/
├── frontend/
│   ├── public/
│   │   ├── index.html
│   │   └── logo.png
│   ├── src/
│   │   ├── App.js              # Main app with landing page
│   │   ├── index.js
│   │   ├── index.css           # Tailwind directives
│   │   └── components/
│   │       ├── FileUpload.js   # VCF upload component
│   │       ├── DrugInput.js    # Drug selection
│   │       └── Results.js      # Analysis results display
│   ├── tailwind.config.js
│   └── package.json
│
├── backend/
│   ├── server.js               # Express server
│   ├── routes/
│   │   └── analyze.js          # API endpoints
│   ├── services/
│   │   ├── phenotypeMapper.js  # CPIC phenotype tables
│   │   ├── riskEngine.js       # Drug risk rules
│   │   └── llmService.js       # OpenAI explanation
│   ├── parser/
│   │   └── vcfParser.js        # VCF file parser
│   ├── .env                    # API keys (not committed)
│   └── package.json
│
├── sample.vcf                  # Test VCF file
└── README.md

🚀 Quick Start

Prerequisites

Node.js 18+
OpenAI API key

1. Clone Repository

git clone https://github.com/yourusername/PharmaGuard.git
cd PharmaGuard

2. Backend Setup

cd backend
npm install

# Create .env file
echo "OPENAI_API_KEY=your_openai_api_key_here" > .env
echo "PORT=3001" >> .env

# Start server
npm start

Backend runs on http://localhost:3001

3. Frontend Setup

cd frontend
npm install
npm start

Frontend runs on http://localhost:3000

📡 API Endpoints

POST `/api/analyze`

Analyze VCF file against specified drugs.

Request: multipart/form-data

vcfFile: VCF file
drugs: Comma-separated drug names (e.g., "CLOPIDOGREL,CODEINE")

Response: Competition-compliant JSON schema

GET `/api/supported-drugs`

Returns list of supported drugs.

POST `/api/validate-vcf`

Validate VCF file without full analysis.

💊 Supported Drugs & Genes

Drug	Gene	Key Variants
CLOPIDOGREL	CYP2C19	rs4244285 (2), rs12248560 (17)
CODEINE	CYP2D6	rs3892097 (4), rs5030655 (6)
WARFARIN	CYP2C9	rs1799853 (2), rs1057910 (3)
SIMVASTATIN	SLCO1B1	rs4149056 (*5)
AZATHIOPRINE	TPMT	rs1800462 (2), rs1142345 (3A)
FLUOROURACIL	DPYD	rs3918290 (*2A), rs67376798

🧬 CPIC Phenotype Classification

CYP2C19 (Clopidogrel)

Diplotype	Phenotype
1/1	Normal Metabolizer (NM)
1/17	Rapid Metabolizer (RM)
17/17	Ultra-rapid Metabolizer (UM)
1/2, 1/3, 2/17**	Intermediate Metabolizer (IM)
2/2, 2/3, 3/3	Poor Metabolizer (PM)

⚠️ Note: *2/*17 is correctly classified as IM (not RM) per CPIC guidelines

CYP2D6 (Codeine)

Diplotype	Phenotype
1/1	NM
1/4	IM
4/4	PM
1/2xN	UM

⚠️ Risk Classification

Risk Label	Severity	Clinical Meaning
🟢 Safe	none	Standard dosing appropriate
🟡 Adjust Dosage	moderate	Dose modification per CPIC
🔴 Toxic	critical	High toxicity risk, avoid or reduce significantly
🔴 Ineffective	high	Therapeutic failure expected
⚪ Unknown	low	Insufficient data

📤 JSON Output Schema

{
  "patient_id": "PATIENT_XXX",
  "drug": "CLOPIDOGREL",
  "timestamp": "2026-02-19T10:30:00.000Z",
  "risk_assessment": {
    "risk_label": "Adjust Dosage",
    "confidence_score": 0.90,
    "severity": "moderate"
  },
  "pharmacogenomic_profile": {
    "primary_gene": "CYP2C19",
    "diplotype": "*2/*17",
    "phenotype": "IM",
    "detected_variants": [
      { "rsid": "rs4244285" },
      { "rsid": "rs12248560" }
    ]
  },
  "clinical_recommendation": {
    "action": "Dose modification recommended...",
    "rationale": "Reduced CYP2C19 function may decrease..."
  },
  "llm_generated_explanation": {
    "summary": "...",
    "mechanism": "...",
    "clinical_impact": "..."
  },
  "quality_metrics": {
    "vcf_parsing_success": true
  }
}

🏗 Architecture

┌─────────────────────────────────────────────────────────────┐
│                      Frontend (React)                        │
│  ┌──────────┐  ┌──────────┐  ┌──────────┐                   │
│  │FileUpload│  │DrugInput │  │ Results  │                   │
│  └────┬─────┘  └────┬─────┘  └────┬─────┘                   │
└───────┼─────────────┼─────────────┼─────────────────────────┘
        │             │             │
        ▼             ▼             ▼
┌─────────────────────────────────────────────────────────────┐
│                    Backend (Express)                         │
│  ┌─────────────────────────────────────────────────────┐    │
│  │                  /api/analyze                        │    │
│  │  ┌──────────┐  ┌────────────┐  ┌─────────────────┐  │    │
│  │  │VCF Parser│→ │Phenotype   │→ │Risk Engine      │  │    │
│  │  │          │  │Mapper      │  │(CPIC Rules)     │  │    │
│  │  └──────────┘  │(CPIC Rules)│  └────────┬────────┘  │    │
│  │                └────────────┘           │           │    │
│  │                                         ▼           │    │
│  │                              ┌─────────────────────┐│    │
│  │                              │ LLM Service        ││    │
│  │                              │ (Explanation Only) ││    │
│  │                              └─────────────────────┘│    │
│  └─────────────────────────────────────────────────────┘    │
└─────────────────────────────────────────────────────────────┘

📖 Usage Examples

Example 1: Analyzing Clopidogrel Risk

Upload a VCF file containing CYP2C19 variants
Select "CLOPIDOGREL" from the drug list
Click "Analyze"

Sample Input:

Variants: rs4244285 (CYP2C19*2), rs12248560 (CYP2C19*17)
Drug: CLOPIDOGREL

Expected Output:

{
  "diplotype": "*2/*17",
  "phenotype": "IM",
  "risk_label": "Adjust Dosage",
  "severity": "moderate"
}

Example 2: Codeine Ultra-Rapid Metabolizer Detection

Sample Input:

Variants: rs1080985 (CYP2D6*2xN duplication)
Drug: CODEINE

Expected Output:

{
  "diplotype": "*1/*2xN",
  "phenotype": "UM",
  "risk_label": "Toxic",
  "severity": "critical"
}

Example 3: Testing via cURL

curl -X POST http://localhost:3001/api/analyze \
  -F "[email protected]" \
  -F "drugs=CLOPIDOGREL,CODEINE"

Example 4: Multiple Drug Analysis

# Analyze multiple drugs at once
curl -X POST http://localhost:3001/api/analyze \
  -F "vcfFile=@patient_data.vcf" \
  -F "drugs=CLOPIDOGREL,WARFARIN,SIMVASTATIN"

👥 Team Members

Name	Role	LinkedIn
Mohit Yadav	Full Stack Developer	LinkedIn Profile
Aditya Kesharwani	Backend Developer	https://www.linkedin.com/in/aditya-keshari-b26903252/
Amogh Patwa	Research & Documentation	https://www.linkedin.com/in/amogh-patwa-926b7525b
Alok Kr. Gupta	Frontend	https://www.linkedin.com/in/alokkgupta28

⚠️ Disclaimer

FOR DEMONSTRATION AND EDUCATIONAL PURPOSES ONLY

This application is a hackathon project. While it implements CPIC-aligned logic, it:

Has NOT been clinically validated
Should NOT be used for actual patient care
Is NOT a substitute for professional medical advice

Always consult qualified healthcare professionals for pharmacogenomic guidance.

📄 License

MIT License - see LICENSE for details.

Built With

Updates

Alok Kumar Gupta started this project — Oct 05, 2026 05:36 PM EDT

Leave feedback in the comments!

Log in or sign up for Devpost to join the conversation.