Khukuri Virtual Lab

AI-powered drug discovery platform with modular architecture.

Features

• Target Discovery: PPI network analysis, target ranking, literature mining
• Molecule Design: AI-powered generation, property optimization, fragment-based design
• Molecular Docking: AutoDock Vina integration, binding site detection, pose analysis
• ADMET Prediction: Drug-likeness, toxicity, pharmacokinetics
• Resistance Prediction: Multi-target strategies, evolution simulation
• Retrosynthesis: Route planning, synthetic accessibility scoring
• Multi-Agent System: AI agents for autonomous drug discovery workflows

Quick Start

Installation

# Install dependencies
pip install -r requirements.txt


# Setup AutoDock Vina (Unix/Linux/macOS)
bash scripts/setup_vina.sh

Quick Test

python scripts/quick_test.py

Run Tests

python -m pytest tests/ -v

Project Structure

khukuri/
├── src/                    # Source code (9 modules, 39 files)
│   ├── core/              # Logging, validation, scoring
│   ├── target_discovery/  # Network analysis, target ranking
│   ├── molecule_design/   # Generation, optimization
│   ├── docking/           # Vina wrapper, pose analysis
│   ├── admet/             # Properties, toxicity, PK/PD
│   ├── resistance/        # Prediction, multi-target
│   ├── synthesis/         # Retrosynthesis, SA scoring
│   ├── agents/            # AI agents, orchestrator
│   └── workflows/         # End-to-end pipelines
├── tests/                 # Test suite (13 files)
├── config/                # Configuration files
├── scripts/               # Automation scripts
└── examples/              # Usage examples

Usage Example

from src.workflows import run_autonomous_discovery

# Run autonomous drug discovery
results = run_autonomous_discovery(
    disease="tuberculosis",
    target_genes=["inhA", "katG"],
    num_candidates=10
)

Dependencies

• RDKit >= 2022.09.1
• NetworkX >= 2.6.0
• BioPython >= 1.79
• NumPy, Pandas, SciPy
• PyYAML >= 6.0
• Requests >= 2.26.0
• python-louvain >= 0.16
• OpenAI >= 1.0.0 (optional)

Note: Work in Progress