bulkrna-de

name	bulkrna-de
description	Bulk RNA-seq differential expression analysis using PyDESeq2 with optional edgeR/limma-voom via rpy2.
version	0.1.0
author	OmicsClaw
license	MIT
tags	["bulkrna","differential-expression","DESeq2","PyDESeq2","edgeR","limma"]
metadata	{"omicsclaw":{"domain":"bulkrna","emoji":"🔬","trigger_keywords":["bulk differential expression","bulk de","deseq2","bulk rna de","differentially expressed genes"]}}
source_reliability	[{"source":"https://github.com/owkin/PyDESeq2","description":"PyDESeq2 main repository maintained by Owkin/scverse; documents supported designs, installation paths, and CI coverage.","score":0.91,"rationale":"Official MIT-licensed implementation with 600+ stars, active releases (0.5.x, Feb 2026), and maintained test matrix for Python 3.11–3.13.\n"},{"source":"https://github.com/K-Dense-AI/claude-scientific-skills","description":"Provides the canonical SKILL.md template and automation guardrails adopted for OmicsClaw skills.","score":0.9,"rationale":"Enterprise-maintained skill pack from K-Dense AI; we mirror its metadata layout to stay interoperable with Claude Scientific Skills.\n"}]

Bulk RNA-seq Differential Expression

Differential expression analysis for bulk RNA-seq count data. Primary engine is PyDESeq2 (a pure-Python re-implementation of DESeq2); falls back to a scipy Welch's t-test with Benjamini-Hochberg FDR correction when PyDESeq2 is not installed.

CLI Reference

python omicsclaw.py run bulkrna-de --demo
python omicsclaw.py run bulkrna-de --input <counts.csv> --output <dir>
python bulkrna_de.py --input counts.csv --output results/ --control-prefix ctrl --treat-prefix treat
python bulkrna_de.py --demo --output /tmp/bulkrna_de_demo
python bulkrna_de.py --input counts.csv --output results/ --method ttest --padj-cutoff 0.01 --lfc-cutoff 1.5

Why This Exists

• Without it: Researchers must manually install and configure DESeq2 in R, handle count normalization, dispersion estimation, and multiple-testing correction across thousands of genes.
• With it: A single Python command runs the full DESeq2 pipeline (via PyDESeq2), produces publication-ready volcano and MA plots, and exports filtered DE tables ready for downstream enrichment.
• Why OmicsClaw: Wraps the gold-standard negative-binomial GLM approach into the OmicsClaw reporting framework with automatic fallback to simpler statistics when dependencies are unavailable.

Workflow

1. Load: Read a genes-by-samples raw count matrix (CSV with a gene column and sample columns prefixed by condition).
2. Partition: Split sample columns into control and treatment groups by prefix matching.
3. Model: Fit a negative-binomial GLM per gene using PyDESeq2 (size-factor normalization, dispersion shrinkage, Wald test). Falls back to Welch's t-test with manual Benjamini-Hochberg FDR if PyDESeq2 is not available.
4. Filter: Identify significant genes at the user-specified padj and log2FC cutoffs.
5. Visualize: Generate volcano plot, MA plot, and DE summary bar chart.
6. Report: Write markdown report, result.json, full and filtered DE tables, and a reproducibility script.

PyDESeq2 Enhancements Adopted

• Multi-factor + continuous covariates: Accept formulas like ~ batch + condition or ~ age + condition, mirroring the design support described in the PyDESeq2 repository (v0.5+) so users can adjust for confounders without leaving Python.
• Installer parity: pip install pydeseq2 (PyPI) or ; OmicsClaw checks for either path at runtime before falling back to the Welch t-test.