Installation

InterScale is available in Python >3.11. All tutorials can be run with CPU support. However, depending on the dataset size, we recommend to train InterScale models on a device with GPU support.

For the fastest installation experience for unimodal training, use the uv package manager within a python-venv environment. For example, run:

python3 -m venv ${/path/to/new/virtual/environment}
source ${/path/to/new/virtual/environment}/bin/activate
pip install uv

where ${/path/to/new/virtual/environment} should be replaced with the path where you want to install the virtual environment.

PyPi

Install InterScale via pip:

uv pip install interscale

Docker container

The CPU supported Docker container can be found here: francescadr/interscale.

The Docker container was set up with viash.

Additional Libraries

To use InterScale, you first need to install some external libraries. These include:

• PyTorch

• PyTorch Scatter

• PyTorch Sparse

• geome

Install all additional dependencies by:

# Create a virtual environment with Python 3.13
uv venv .interscale --python 3.13
source .interscale/bin/activate

# Install PyTorch (CPU/MPS build for macOS)
uv pip install torch torchvision torchaudio

# PyG extensions — use the CPU wheel index
uv pip install torch-scatter torch-sparse torch-cluster \
  -f https://data.pyg.org/whl/torch-2.10.0+cpu.html

# Core dependencies
uv pip install torch-geometric pytorch-lightning wandb yacs scvi-tools
uv pip install geome