Installation¶

Requirements¶

NeKo requires Python ≥ 3.10.

Install from PyPI (recommended)¶

pip install nekomata

Package name

The PyPI package is called nekomata — pip install neko and pip install pyneko are unrelated packages.

Install from source¶

For the latest development version:

git clone https://github.com/sysbio-curie/Neko.git
cd Neko
pip install .

Optional: install with Poetry¶

git clone https://github.com/sysbio-curie/Neko.git
cd Neko
poetry install

System dependencies¶

Graphviz¶

Graphviz is required for network rendering. Install it with your system package manager before installing NeKo.

Linux (apt)macOS (Homebrew)Windows (Chocolatey)

sudo apt-get install python3-dev graphviz libgraphviz-dev

brew install graphviz

choco install graphviz

More information: https://graphviz.org/download/

Pandoc (optional, for notebook export)¶

sudo apt-get install pandoc   # Linux
brew install pandoc           # macOS

Verify installation¶

import neko
print(neko.__version__)

Troubleshooting¶

`graphviz` import errors¶

If import graphviz fails, ensure the system Graphviz binary is on your PATH:

which dot   # should point to the graphviz dot binary

OmniPath and UniProt connectivity¶

NeKo downloads interaction data from OmniPath when building an OmniPath universe. Identifier translation lazily downloads a reviewed-human mapping table from UniProt and caches it locally. SIGNOR similarly caches its validated interaction table and entity dictionaries, so later loads can run offline. If a resource contains ChEBI accessions, NeKo also caches the compressed ChEBI compounds table and a small accession-to-name subset. The names are optional: if ChEBI is unreachable, the canonical accession is used as the display label and network construction continues. Ensure the first use has network access, or provide an existing cache with NEKO_CACHE_DIR.

from neko.inputs import Universe

resources = Universe("omnipath")