HIV Drug Prediction Model

A machine learning model to predict which chemical compounds can fight HIV effectively, helping researchers focus on the most promising candidates.

Overview

This project uses QSAR (Quantitative Structure-Activity Relationship) modeling to predict HIV drug compound efficacy. It helps pharmaceutical researchers identify promising compounds before expensive lab testing.

Key Benefits:

• Reduce screening time from months to hours
• Focus resources on high-probability compounds
• Improve success rates in drug discovery

Features

• Complete ML pipeline from data preprocessing to deployment
• REST API for real-time predictions
• Model monitoring and performance tracking
• Batch processing for large compound libraries
• Docker containerization for easy deployment

Dataset

Source: NCI AIDS Antiviral Screen Data

• 40,000+ HIV-tested compounds
• Activity classes: CA (Active), CM (Moderately Active), CI (Inactive)
• EC50/IC50 measurements
• Molecular structure data

Technology Stack

• Language: Python 3.8+
• ML Libraries: scikit-learn, XGBoost, RDKit
• API: FastAPI
• Database: PostgreSQL
• MLOps: MLflow
• Deployment: Docker

Model Performance

Metric	Value
Accuracy	87.3%
F1-Score	0.84
Cohen's Kappa	0.79
AUC-ROC	0.91

Quick Start

Installation

# Clone repository
git clone https://github.com/pari1jay/Prediction-Model-HC.git
cd Prediction-Model-HC

# Create virtual environment
python -m venv venv
source venv/bin/activate  # Linux/Mac
# venv\Scripts\activate   # Windows

# Install dependencies
pip install -r requirements.txt

Running the Application

# Development mode
python src/main.py

# Production with Docker
docker-compose up -d

API Usage

import requests

# Predict compound activity
response = requests.post(
    "http://localhost:8000/predict",
    json={"smiles": "CC(C)CC1=CC=C(C=C1)C(C)C(=O)O"}
)
print(response.json())

Usage Examples

Training a Model

from src.training.pipeline import TrainingPipeline

pipeline = TrainingPipeline()
pipeline.load_data("data/hiv_compounds.csv")
pipeline.preprocess()
pipeline.train()
pipeline.evaluate()

Making Predictions

from src.prediction.predictor import CompoundPredictor

predictor = CompoundPredictor.load("models/best_model.pkl")
result = predictor.predict_smiles("CCO")
print(f"Activity: {result['activity']}, Confidence: {result['confidence']:.3f}")

Batch Processing

python scripts/batch_predict.py --input compounds.csv --output predictions.csv

Project Structure

Prediction-Model-HC/
├── src/
│   ├── data/           # Data processing
│   ├── features/       # Feature engineering
│   ├── models/         # ML models
│   ├── training/       # Training pipeline
│   ├── prediction/     # Prediction service
│   └── api/           # REST API
├── data/              # Datasets
├── models/            # Trained models
├── scripts/           # Utility scripts
├── tests/             # Test files
└── docs/              # Documentation

Data Processing Pipeline

1. Data Integration: Merge screening results, EC50/IC50 values, and molecular structures
2. Quality Control: Handle duplicates, conflicts, and missing data
3. Feature Engineering: Calculate molecular descriptors and fingerprints
4. Model Training: Train and validate multiple ML models
5. Evaluation: Assess model performance using relevant metrics

Contributing

1. Fork the repository
2. Create a feature branch (git checkout -b feature/new-feature)
3. Commit your changes (git commit -m 'Add new feature')
4. Push to the branch (git push origin feature/new-feature)
5. Open a Pull Request

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgments

• National Cancer Institute for the AIDS Antiviral Screen Data
• RDKit community for cheminformatics tools
• Open source contributors

Contact

Pari Jay - GitHub

Project Link: https://github.com/pari1jay/Prediction-Model-HC