Torchdrug

This is your go-to for building graph neural networks on molecules and proteins with PyTorch. It comes with 40+ datasets (BBBP, Tox21, QM9 for molecules; EnzymeCommission, PDBBind for proteins), 20+ model architectures (GIN, GAT, SchNet, GearNet), and task wrappers for property prediction, retrosynthesis, and knowledge graph reasoning. The installation is a bit finicky since you need to match PyTorch versions with torch-scatter and torch-cluster wheels, but once that's sorted you get a complete drug discovery stack. It's genuinely comprehensive but last updated July 2023, so you're working with PyTorch 2.0 max and Python 3.10. Best for custom model development where you need full control over the architecture rather than just running inference on pre-trained models.

skills.sh installs: 2K
GitHub stars: 56.6K
last updated: Aug 21, 2026

View on GitHub

Install

$ npx -y skills add k-dense-ai/scientific-agent-skills --skill torchdrug --agent claude-code

Installs into .claude/skills of the current project.

Files

TorchDrug

Use TorchDrug as a modular PyTorch graph-learning stack:

load a datasets.* dataset,
choose a models.* representation model,
wrap it in a tasks.* objective,
train and evaluate it with core.Engine.

The current official documentation and latest published release are both 0.2.1 (released July 2023; rechecked October 1, 2026). Treat newer Python or PyTorch combinations as unverified rather than silently assuming compatibility.

Start with the version guard

Before generating or debugging code, inspect the environment:

python --version
python -c "import torch; print(torch.__version__)"
python -c "import torchdrug; print(torchdrug.__version__)"

The supported matrix for TorchDrug 0.2.1 is:

Python 3.7 through 3.10
PyTorch 1.8 through 2.0
Linux, Windows, or macOS
Apple Silicon: PyTorch 1.13 or later, CPU only; no MPS support

If the project uses Python 3.11+ or PyTorch 2.1+, create a compatible environment or explicitly test a source build. Do not present such combinations as supported.