ederiv2 0.1.0

A molecular graph generation and analysis toolkit using Graph Neural Networks

eDeriv2

A molecular graph generation and analysis toolkit using Graph Neural Networks for drug discovery and molecular design.

Overview

eDeriv2 is a comprehensive Python package for molecular graph generation, analysis, and machine learning applications in chemistry and drug discovery. It provides state-of-the-art Graph Neural Network (GNN) models for molecular representation learning, graph generation, and molecular property prediction.

Features

• Molecular Graph Generation: Advanced GNN-based models for generating molecular graphs
• Graph Neural Networks: Implementation of various GNN architectures (GVAE, GAE, EGATConv)
• Molecular Analysis: Tools for molecular property prediction and analysis
• RDKit Integration: Seamless integration with RDKit for molecular operations
• DGL Support: Built on Deep Graph Library (DGL) for efficient graph operations
• PyTorch Backend: Full PyTorch support for deep learning models
• Visualization: Built-in visualization tools for molecular graphs and results

Installation

From PyPI (Recommended)

pip install ederiv2

From Source

# Clone the repository
git clone https://github.com/yourusername/eDeriv2.git
cd eDeriv2

# Install in development mode
pip install -e .

Dependencies

The package requires the following key dependencies:

• Python >= 3.8
• PyTorch >= 1.9.0
• DGL >= 1.0.0
• RDKit >= 2022.9.1
• NumPy >= 1.21.0
• Pandas >= 1.3.0

For a complete list of dependencies, see requirements.txt.

Quick Start

Basic Usage

import torch
from src.graph_handler import DGLGraphHandler
from src.gvae_models import GVAE

# Initialize a GVAE model
model = GVAE(
    node_feat_dim=13,
    edge_feat_dim=4,
    hidden_dim=64,
    latent_dim=32,
    node_classes=13,
    edge_classes=4
)

# Create a graph handler
handler = DGLGraphHandler()

# Your molecular data processing here
# ...

Molecular Graph Generation

from src.graph_maker import DGLGraphMaker
from rdkit import Chem

# Create a graph maker
graph_maker = DGLGraphMaker()

# Convert SMILES to graph
smiles = "CCO"
mol = Chem.MolFromSmiles(smiles)
graph = graph_maker.create(mol, "rdkit_mol")

Training a Model

from src.nn_tools.trainers import GVAETrainer

# Initialize trainer
trainer = GVAETrainer(model, device='cuda')

# Train the model
trainer.train(train_dataloader, val_dataloader, epochs=100)

Project Structure

eDeriv2/
├── src/                          # Main package source
│   ├── chem_handlers/           # Chemical data handling
│   ├── input_tools/             # Input processing tools
│   ├── nn_tools/                # Neural network utilities
│   ├── optm_tools/              # Optimization tools
│   ├── output_tools/            # Output and visualization
│   └── sys_tools/               # System utilities
├── assets/                      # Data assets
├── outputs/                     # Output files
├── training_plots/              # Training visualizations
├── setup.py                     # Package setup
├── pyproject.toml              # Modern Python packaging
├── requirements.txt            # Dependencies
└── README.md                   # This file

Models

GVAE (Graph Variational Autoencoder)

• File: gvae_v1.py, gvae_v2.py
• Description: Graph Variational Autoencoder for molecular graph generation
• Features: Encoder-decoder architecture with variational inference

GAE (Graph Autoencoder)

• File: gae.py
• Description: Graph Autoencoder for graph representation learning
• Features: Simple autoencoder architecture for graphs

EGATConv (Edge-aware Graph Attention)

• File: graph_encoder.py
• Description: Edge-aware Graph Attention Convolution
• Features: Attention mechanism for both nodes and edges

Examples

Molecular Property Prediction

from src.models import MolecularPropertyPredictor

# Initialize predictor
predictor = MolecularPropertyPredictor(model_path="path/to/model.pth")

# Predict properties
properties = predictor.predict(smiles_list)

Graph Visualization

from src.utils import plot_molecules_and_fragments

# Visualize molecular graphs
plot_molecules_and_fragments(molecules, fragments, output_path="output.png")

Contributing

We welcome contributions! Please see our Contributing Guidelines for details.

Development Setup

# Clone the repository
git clone https://github.com/yourusername/eDeriv2.git
cd eDeriv2

# Create a virtual environment
python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

# Install development dependencies
pip install -e ".[dev]"

# Run tests
pytest

License

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

Citation

If you use eDeriv2 in your research, please cite:

@software{ederiv2,
  title={eDeriv2: A molecular graph generation and analysis toolkit},
  author={eDeriv2 Team},
  year={2024},
  url={https://github.com/yourusername/eDeriv2}
}

Support

• Documentation: https://github.com/yourusername/eDeriv2#readme
• Issues: https://github.com/yourusername/eDeriv2/issues
• Discussions: https://github.com/yourusername/eDeriv2/discussions

Acknowledgments

• RDKit for molecular informatics
• DGL for deep graph library
• PyTorch for deep learning framework