Installation¶
pip install "spatial-smooth[all]"
Note
[all] pulls every optional backend. For a minimal install, drop the extra
(pip install spatial-smooth) and add only the extras you need (see below). To install the
development version straight from the repository:
pip install "spatial-smooth[all] @ git+https://github.com/settylab/spatial-smooth".
Only numpy, scipy, pandas and anndata are hard requirements. Everything else is
an extra, imported lazily, and reported with the exact pip install line when missing.
extra |
brings |
needed for |
|---|---|---|
|
the Gaussian-process step ( |
|
|
computing the diffusion map |
|
|
|
plotting |
|
|
the squidpy plotting backend |
|
|
the fine-grid FFT smoother ( |
Check where you stand at any time:
import spatial_smooth as ss
ss.check_dependencies()
spatial_smooth v0.1.0 -- dependency check
numpy 2.4.6 ok
pandas 2.3.3 ok
scipy 1.18.0 ok
anndata 0.12.17 ok
kompot - missing: pip install "kompot>=0.7.0" (optional)
...
Every missing optional dependency raises where it is first needed, naming the package, what it
is for, and how to install it – never a bare ModuleNotFoundError from deep in a call stack.
Reading a smoothed object needs none of the extras: a colleague can open your .h5ad and
plot it with scanpy alone.