SpatialDE Modern

spatialde-modern is a maintained fork of SpatialDE, a method for identifying genes whose expression depends on spatial or temporal coordinates.

The distribution keeps the original SpatialDE import package and numerical API while supporting current Python, NumPy, pandas, and SciPy releases. It is published under a distinct distribution name so it cannot be confused with an official release from the original authors.

Installation

pip install spatialde-modern

The import name remains unchanged:

import SpatialDE

Minimal example

import pandas as pd
import SpatialDE

coordinates = pd.DataFrame({"x": [0, 0, 1, 1], "y": [0, 1, 0, 1]})
expression = pd.DataFrame(
    {
        "gene_a": [0.1, 0.2, 1.1, 1.2],
        "gene_b": [0.5, 0.4, 0.6, 0.5],
    }
)

results = SpatialDE.run(coordinates, expression)

Maintenance scope

This repository contains only the maintained Python distribution. The original paper analyses, datasets, notebooks, R implementation, and historical Stan experiments remain available in the upstream repository. They are not runtime dependencies and are intentionally excluded here.

Compatibility fixes in this fork include:

• replacement of the removed scipy.misc.derivative API;
• conversion of removed top-level SciPy array aliases to NumPy;
• support for current pandas array semantics;
• built-in stabilize and regress_out preprocessing, replacing the separate NaiveDE runtime dependency used by the original command-line workflow;
• modern PEP 517/621 packaging with complete runtime dependencies;
• isolated wheel builds and real numerical smoke tests in CI.

Attribution and license

SpatialDE was created by Valentine Svensson and collaborators. This maintained fork preserves the original MIT license and records its upstream base in NOTICE.md.