Installation¶
antibody-utils requires Python 3.11 or later.
With uv¶
Add antibody-utils to a uv project:
uv add antibody-utils
With pip¶
pip install antibody-utils
Optional extras¶
numberingNumber raw sequences and structures with ANARCII. This pulls in PyTorch, which is a large download.
uv add "antibody-utils[numbering]"pip install "antibody-utils[numbering]"Tip
On Linux, the default PyTorch wheels from PyPI bundle CUDA libraries, a download of several gigabytes. To get the build that suits your machine (CPU-only, or your CUDA or ROCm version):
With uv’s pip interface, let uv detect your hardware:
uv pip install --torch-backend=auto "antibody-utils[numbering]"uv addhas no such flag; in a uv project, pointtorchat the right PyTorch index in yourpyproject.tomlinstead, as described in uv’s PyTorch guide.pip cannot detect your hardware, so install PyTorch first from the PyTorch index for your platform, e.g. CPU-only:
pip install torch --index-url https://download.pytorch.org/whl/cpu pip install "antibody-utils[numbering]"
Installing for development¶
Clone the repository, fetch the BLOSUM62 matrix (not tracked in git) and install the development environment with uv:
git clone https://github.com/oxpig/antibody-utils.git
cd antibody-utils
uv run --no-project python scripts/fetch_blosum62.py
uv sync
uv run pytest