Install gpgraph-v2¶

gpgraph-v2 ships prebuilt wheels for Linux (x86_64, aarch64), macOS (x86_64, aarch64), and Windows (x64). Python 3.11 or newer is required.

Stable release¶

uv add gpgraph-v2

pip install gpgraph-v2

The package installs the import path gpgraph, regardless of the distribution name being gpgraph-v2.

import gpgraph
print(gpgraph.__version__)

Optional extras¶

Matplotlib is split out so the core install stays headless.

uv add "gpgraph-v2[plot]"

pip install "gpgraph-v2[plot]"

Importing gpgraph.pyplot without matplotlib installed raises a clear ImportError directing you to install the extra.

Build from source¶

You need a Rust toolchain (stable, >= 1.70) and maturin. The recommended workflow uses uv:

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

After editing Rust under crates/:

uv run maturin develop --release && uv run pytest

Consuming from another local project¶

For co-development with sister packages like gpmap-v2 and epistasis-v2, point at a local editable copy in your consumer's pyproject.toml:

[tool.uv.sources]
gpgraph-v2 = { path = "/absolute/path/to/gpgraph-v2", editable = true }
gpmap-v2 = { path = "/absolute/path/to/gpmap-v2", editable = true }

[project]
dependencies = ["gpgraph-v2", "gpmap-v2"]

Then uv sync in the consumer. Imports remain from gpgraph import GenotypePhenotypeGraph.

Verify the install¶

from gpmap import GenotypePhenotypeMap
from gpgraph import GenotypePhenotypeGraph

gpm = GenotypePhenotypeMap(
    wildtype="AA",
    genotypes=["AA", "AT", "TA", "TT"],
    phenotypes=[0.1, 0.2, 0.3, 1.0],
)
G = GenotypePhenotypeGraph.from_gpm(gpm)
print(G.number_of_nodes(), G.number_of_edges())  # 4 8

If G reports 4 nodes and 8 directed edges, the Rust extension is wired in correctly.