naivede-modern

naivede-modern is a maintained fork of the original NaiveDE project. It keeps the public NaiveDE import package and numerical API while providing current packaging, complete runtime dependency metadata, and support for Python 3.11 through 3.14. The original authors and MIT license are preserved.

Install the maintained distribution with:

pip install naivede-modern

This package implements the simplest form of a Differential Expression test. It just fits two linear models, assuming Normally distributed residuals, to each gene. These are assumed to be nested, and a likelihood ratio test is then performed to compare the models.

The test might have low power and give noisy results, but it shouldn't be biased. With many (i.e. hundreds of) samples it should work all right.

This can serve as a baseline comparison with more sophisticated tests.

The optional cell-type classification helpers require scikit-learn:

pip install "naivede-modern[cell-types]"

Faux fold changes

The package also has methods for creating input fold-change controlled fake conditions in data using ERCC spike-ins. This performs systematic renaming of spike-ins in randomized conditions. To avoid creating unrealistic levels of fold change, input concentration is used to limit possible renaming.

This assumes expression measures used are comparable between different sequences. Thus these should optimally account for length and other sequence features before creating faux fold changes.

Maintenance scope

This repository contains only the maintained Python distribution. The original analysis notebook remains available in the upstream repository; it is not a runtime dependency and is intentionally excluded here.

Attribution and license

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