BIOFLOW_README.md

BioFlow - AI-Powered Drug Discovery Platform

BioFlow is a unified AI platform for drug discovery, combining molecular encoding, protein analysis, and drug-target interaction prediction in a modern web interface.

🏗️ Architecture

┌─────────────────────────────────────────────────────────────┐
│                    Next.js Frontend                          │
│                   (React 19 + Tailwind)                      │
│                     localhost:3000                           │
└───────────────────────┬─────────────────────────────────────┘
                        │ HTTP/REST
                        ▼
┌─────────────────────────────────────────────────────────────┐
│                   FastAPI Backend                            │
│                    localhost:8000                            │
│  ┌──────────────┐  ┌─────────────┐  ┌──────────────────┐   │
│  │ ModelService │  │QdrantService│  │ DTI Predictor    │   │
│  │ (Encoders)   │  │ (VectorDB)  │  │ (DeepPurpose)    │   │
│  └──────────────┘  └─────────────┘  └──────────────────┘   │
└───────────────────────┬─────────────────────────────────────┘
                        │
                        ▼
┌─────────────────────────────────────────────────────────────┐
│                    OpenBioMed Core                           │
│  ┌────────────┐  ┌────────────┐  ┌─────────────────────┐   │
│  │   Models   │  │  Datasets  │  │       Tasks         │   │
│  │ BioT5,ESM  │  │ DAVIS,KIBA │  │ Property Prediction │   │
│  └────────────┘  └────────────┘  └─────────────────────┘   │
└─────────────────────────────────────────────────────────────┘

🚀 Quick Start

Prerequisites

• Python 3.10+
• Node.js 18+ with pnpm
• (Optional) CUDA-compatible GPU

Installation

# Clone the repository
git clone https://github.com/hamzasammoud11-dotcom/lacoste001.git
cd lacoste001

# Install Python dependencies
pip install -r bioflow/api/requirements.txt

# Install frontend dependencies
cd lacoste001/ui
pnpm install
cd ../..

Running

Option 1: Using the launch script (Windows)

launch_bioflow_full.bat

Option 2: Manual start

# Terminal 1: Start FastAPI backend
python -m uvicorn bioflow.api.server:app --reload --port 8000

# Terminal 2: Start Next.js frontend
cd lacoste001/ui
pnpm dev

Access

• Frontend: http://localhost:3000
• API Docs: http://localhost:8000/docs
• API Health: http://localhost:8000/health

📁 Project Structure

OpenBioMed/
├── bioflow/                    # BioFlow Platform
│   ├── api/                    # FastAPI Backend
│   │   ├── server.py           # Main API server
│   │   ├── model_service.py    # Unified model access
│   │   ├── qdrant_service.py   # Vector database
│   │   └── dti_predictor.py    # DTI prediction
│   ├── core/                   # Core abstractions
│   ├── plugins/                # Encoders & retrievers
│   └── workflows/              # Pipeline definitions
│
├── lacoste001/
│   └── ui/                     # Next.js Frontend
│       ├── app/
│       │   ├── api/            # API routes
│       │   └── dashboard/      # UI pages
│       ├── components/         # React components
│       └── lib/                # Services & utilities
│
├── open_biomed/                # OpenBioMed Research Engine
│   ├── models/                 # BioT5, ESM, GraphMVP
│   ├── datasets/               # Dataset loaders
│   └── tasks/                  # Task implementations
│
└── configs/                    # YAML configurations

🔌 API Endpoints

Discovery Pipeline

• POST /api/discovery - Start discovery job
• GET /api/discovery/{job_id} - Get job status

Predictions

• POST /api/predict - DTI prediction
• POST /api/encode - Encode molecule/protein/text

Data Management

• POST /api/ingest - Add data to vector DB
• GET /api/molecules - List molecules
• GET /api/proteins - List proteins
• GET /api/collections - List vector collections

Visualization

• GET /api/explorer/embeddings - Get 2D projections
• GET /api/similarity - Compute similarity scores

🧪 Features

Drug Discovery Pipeline

• Natural language, SMILES, or FASTA input
• Automatic modality detection
• Vector similarity search
• Property prediction (MW, LogP, TPSA)
• Binding affinity prediction

Molecular Analysis

• 2D/3D molecule visualization
• SMILES validation
• Property calculation via RDKit

Protein Analysis

• 3D protein structure viewing
• Sequence embedding
• DTI prediction

Explorer

• UMAP/t-SNE embedding visualization
• Cluster analysis
• Interactive filtering

🔧 Configuration

Environment Variables

# .env file
NEXT_PUBLIC_API_URL=http://localhost:8000
QDRANT_URL=http://localhost:6333  # Optional: remote Qdrant
QDRANT_PATH=./qdrant_data          # Local Qdrant storage

API Configuration

Edit lacoste001/ui/config/api.config.ts:

export const API_CONFIG = {
  baseUrl: process.env.NEXT_PUBLIC_API_URL || "http://localhost:8000",
  // ...
}

🧬 Model Support

Model	Type	Use Case
ChemBERTa	Molecule Encoder	SMILES embeddings
ESM-2	Protein Encoder	Sequence embeddings
PubMedBERT	Text Encoder	Biomedical text
DeepPurpose	DTI	Binding prediction
GraphMVP	Property	Molecular properties
BioT5	Generation	Molecule generation

📊 Development

Verify Installation

python scripts/verify_phase3.py

Run Tests

pytest tests/

Type Checking (Frontend)

cd lacoste001/ui
pnpm tsc --noEmit

🤝 Contributing

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Submit a pull request

📄 License

Apache 2.0 - See LICENSE

🙏 Acknowledgments

• OpenBioMed - Foundation models
• DeepPurpose - DTI prediction
• Qdrant - Vector database
• Next.js - React framework
• Shadcn/ui - UI components