Skip to content

PyTorch Geometric

PyTorch Geometric (PyG) is a general graph neural network library with an official QM9 example for molecular property learning.

Catalog updated ·

GitHub / Repository ↗Documentation ↗
Follow

Public link：https://chemaiatlas.com/open-source/pytorch-geometric/

Claim this resource

Overview

It provides graph data and message-passing components across domains. Here it is foundational software for molecular graph research, not a chemistry-specific pretrained model.

Limitations

Models, data and training settings must be selected by the user. Optional accelerators must match the PyTorch and hardware environment.

Key Features

  • Graph data and message passing
  • Official QM9 example

Use Cases

  • Develop molecular graph predictors
  • Reproduce a QM9 learning workflow

How to Use

Install using the official guide and inspect the QM9 example before preparing data and training settings. Record dependencies and evaluation splits.

Related resources

ChemBERTa

Model

ChemBERTa provides BERT-like models for chemical SMILES, with RoBERTa masked-language modelling checkpoints and notebooks for pre-training, fine-tuning and molecular property prediction research.

Open sourcePython

Molecular Property Prediction

Updated

ChemML

Open Source

ChemML is a Python suite for chemical and materials data analysis, mining, and modeling, with a modular design and documented work on graph neural networks, AutoML, and explainability.

Open sourcePython

Molecular Property Prediction · Materials Discovery

Updated

Chemprop

Model

Chemprop is a PyTorch-based toolkit for training and evaluating message-passing neural networks for molecular property prediction, with CLI workflows, Python modules and task-specific notebooks.

Open sourcePython

Molecular Property Prediction

Updated

DeepMol

Open Source

DeepMol is a Python toolkit for molecular machine learning, connecting compound standardization and featurization with model training, evaluation, interpretation and reusable pipelines.

Open sourcePython

Molecular Property Prediction

Updated

DGL-LifeSci

Open Source

DGL-LifeSci is a DGL/PyTorch toolkit providing graph learning components and examples for chemistry and biology.

Open sourcePython

Molecular Property Prediction

Updated

DimeNet

Model

DimeNet provides reference implementations of DimeNet and DimeNet++ for directional message passing on molecular graphs, with training notebooks, test-set prediction workflows and pretrained models.

Python

Molecular Property Prediction · Quantum Chemistry

Updated

Works with

Used in Recipes

Related guides