limmapy

A pure-Python port of Bioconductor limma (Ritchie et al., Nucleic Acids Research 2015) — linear models and empirical-Bayes moderated statistics for differential expression.

• No rpy2, no R install — the core limma linear-model workflow reimplemented in NumPy / SciPy
• The canonical pipeline: voom → lmFit → contrasts.fit → eBayes → topTable
• Smyth 2004 empirical-Bayes variance moderation; treat, decideTests, duplicateCorrelation, removeBatchEffect
• Both Python-style (lm_fit, ebayes, top_table) and R-style (lmFit, eBayes, topTable) names exported
• pandas / numpy-friendly

The import name is pylimma; the PyPI distribution name is python-limma (pip install python-limma). The names pylimma / limmapy were unavailable on PyPI, so the distribution carries the omicverse-ecosystem name.

This is a standalone mirror of the implementation developed in omicverse. It powers omicverse's edgeR / limma-voom differential-expression backends and the pydeqms proteomics workflow.

Install

pip install python-limma

Quick start

import numpy as np
import pylimma

# expr: genes x samples log-expression matrix; design: samples x coefficients
fit = pylimma.lmFit(expr, design)
fit = pylimma.eBayes(fit)
res = pylimma.topTable(fit, coef=1, number=np.inf)
res.head()

RNA-seq with voom

import pylimma
v = pylimma.voom(counts, design)            # EList: mean-variance weights
fit = pylimma.eBayes(pylimma.lmFit(v.E, design, weights=v.weights))
pylimma.topTable(fit, coef=1)

API

Citation

Ritchie, M.E., Phipson, B., Wu, D., Hu, Y., Law, C.W., Shi, W., Smyth, G.K. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Research 43(7), e47 (2015).

…and acknowledge omicverse / this repo for the Python port.

License

LGPL-3.0-or-later — matches the upstream Bioconductor package.