wrapshap 0.1

This package allows you to perform SHAP-based feature selection and exploration for high-dimensional data, with both CPU and GPU support.

Paper: Furger et al. (2026), Wrapshap: a high dimension feature selection framework to explore biological systems through machine learning explanations.

How to install Wrapshap

In order to install the CPU implementation of Wrapshap, simply run:

pip install wrapshap

If you can afford it, the GPU implementation is much faster and comes with additional methods. To install it instead, please run:

pip install "wrapshap[gpu]"

You should now be able to import wrapshap utilities.

Warning: If you are encountering import issues, please try using Python version 3.9

For the CPU implementation:

from wrapshap import Wrapshap

or for the already wrapped models:

from wrapshap.already_wrapped import WrappedBinaryClassifierXGB, WrappedRegressorXGB

For the GPU implementation:

from wrapshap.gpu_implementation import WrapshapGPU

or for the already wrapped models:

from wrapshap.gpu_implementation.already_wrapped_gpu import WrappedRegressorXGB_GPU, WrappedBinaryClassifierXGB_GPU

For more details about how to use the package, please head over to our GitHub repository for example notebooks.