Developer installation
Use either uv (recommended) or Conda. You need only one of them.
Recommended setup
Install uv and prek (a fast pre-commit), then:
uv tool install prek
git clone git@github.com:neuroseq/neuroseq.git
cd neuroseq
uv sync --group dev
prek install
uv sync creates .venv from pyproject.toml and uv.lock; uv run uses it
without activation. Run uv sync again after pulling. To upgrade dependencies,
run uv lock --upgrade, then uv sync, and test (this changes uv.lock).
Alternative setup
git clone git@github.com:neuroseq/neuroseq.git
cd neuroseq
conda create -n neuroseq python=3.12
conda activate neuroseq
python -m pip install --upgrade pip
python -m pip install -e . --group dev
python -m pip install prek
prek install
--group needs a recent pip. Pip ignores uv.lock and [tool.uv.sources], so
package versions and PyTorch builds can differ from the uv setup. Drop uv run
from the commands in these docs (for example python -m pytest).
!!! warning "uv inside Conda"
If uv is installed in an active Conda environment, `uv run` still uses the
repository's `.venv`, not Conda's Python. `conda run -n neuroseq python` and
`uv run python` can then load different packages.
Check which interpreter and versions you are using (point your IDE at the same one):
uv run python -c "import sys, brian2, numpy; print(sys.executable, brian2.__version__, numpy.__version__)"
In the embedded console, import sys; print(sys.executable) shows the running interpreter.
Everyday commands
| Task | Command |
|---|---|
| Launch the GUI | uv run neuroseq |
| Debug logging | uv run neuroseq --log-level DEBUG |
| Run tests | uv run pytest (see Testing) |
| Run all hooks | prek run -a |
| Serve docs | uv run zensical serve (see Documentation) |
| Profile the GUI | uv run python -m cProfile -o latest.prof -m neuroseq, then uvx snakeviz latest.prof |