Install epistasis-v2¶

epistasis-v2 is distributed as a source distribution and a set of platform wheels on PyPI. The wheels bundle pre-compiled Rust extensions, so most users can install the package with a single pip command and no Rust toolchain. This page covers both paths and lists every runtime dependency.

Prerequisites¶

Python 3.10 or later. Supported versions: 3.10, 3.11, 3.12, 3.13.
pip (any recent version). No Rust toolchain is needed when installing from a PyPI wheel.

PyPI installBuild from source

Install the latest release from PyPI:

pip install epistasis-v2

This pulls a pre-built wheel for your platform. The Rust extension (epistasis._core) is already compiled inside the wheel, so the install completes without a Rust toolchain.

Verify the installation¶

Open a Python interpreter and confirm the package is importable:

import epistasis
print(epistasis.__version__)

Tip

If you are working inside a virtual environment or conda environment, activate it before running pip install.

Building from source lets you work with the latest unreleased code or contribute to the library. You need:

A Rust toolchain (stable channel). Install via rustup.rs.
uv for dependency management. Install via pip install uv or the uv docs.

Clone the repository and build:

git clone https://github.com/lperezmo/epistasis-v2
cd epistasis-v2
uv sync
uv run maturin develop --release

uv sync installs all Python dependencies (including dev extras). maturin develop --release compiles the Rust crate and links it into the active environment in release mode.

Verify the build¶

uv run python -c "import epistasis; print(epistasis.__version__)"

Warning

The --release flag is important. Without it, the Rust kernels compile in debug mode and run significantly slower.

Runtime dependencies¶

All dependencies below are installed automatically when you run pip install epistasis-v2.

Package	Minimum version	Purpose
`numpy`	1.23	Array operations and matrix algebra
`pandas`	2.0	Tabular storage for the `EpistasisMap`
`scipy`	1.11	Statistical distributions and linear algebra utilities
`scikit-learn`	1.3	Underlying estimators for linear and regularized models
`lmfit`	1.2	Parameter fitting for nonlinear models
`emcee`	3.1	MCMC ensemble sampler for Bayesian coefficient estimation
`matplotlib`	3.7	Plotting utilities
`gpmap-v2`	1.0.0	`GenotypePhenotypeMap` objects and encoding tables

Note

gpmap-v2 is the companion library that provides GenotypePhenotypeMap. epistasis-v2 models accept GPM objects directly through add_gpm(). The two libraries are versioned and released together; install a matching pair to avoid compatibility issues.

Python version support¶

epistasis-v2 is tested on Python 3.10, 3.11, 3.12, and 3.13 in CI. Python versions older than 3.10 are not supported.

Upgrading¶

To upgrade to the latest release:

pip install --upgrade epistasis-v2