rnaseq-toolkit 0.1.0

Streamlined differential expression analysis and pathway enrichment visualization from RNA-seq data

rnaseq-toolkit

rnaseq-toolkit is an open-source Python package that provides a unified, switchable interface for RNA-seq differential expression analysis (DEA) and pathway enrichment visualization. It automates the complete pipeline from raw count matrices through normalization, DEA, and pathway analysis to self-contained HTML reports — all with a single command or a few lines of Python.

Key Features

• Unified normalization interface: switch between DESeq2 median-of-ratios, TMM, CPM, TPM, RPKM, VST, and rlog with one argument (--norm-method)
• Switchable DEA methods: DESeq2 (via PyDESeq2) and edgeR-like quasi-likelihood (via statsmodels NB GLM) with identical output format
• Pathway enrichment: pre-ranked GSEA, GO over-representation, and KEGG pathway analysis via gseapy/Enrichr
• Publication-quality plots: volcano, MA, PCA, clustered heatmap, GSEA dotplot — all at 300 DPI
• Automated HTML reports: self-contained reports with embedded plots and interactive tables
• Snakemake workflow: reproducible, scalable pipeline management
• Docker support: fully containerized for reproducibility
• Benchmarked: validated on three public GEO datasets (Batten disease, COVID-19, breast cancer)

Installation

pip install rnaseq_toolkit

Or install from source:

git clone https://github.com/rnaseq-toolkit/rnaseq-toolkit.git
cd rnaseq-toolkit
pip install -e ".[dev]"

Quick Start

Command-line interface

# Basic DESeq2 analysis
rnaseq-toolkit \
    --counts data/counts.csv \
    --metadata data/metadata.csv \
    --design "~condition" \
    --contrast condition treated control \
    --output results/

# Switch to edgeR-like with TMM normalization — one argument change
rnaseq-toolkit \
    --counts data/counts.csv \
    --metadata data/metadata.csv \
    --design "~condition" \
    --norm-method tmm \
    --dea-method edger \
    --output results_edger/

Python API

from rnaseq_toolkit import RNAseqPipeline

# Initialize pipeline
pipe = RNAseqPipeline(
    counts_path="data/counts.csv",
    metadata_path="data/metadata.csv",
    design="~condition",
    output_dir="results/",
)

# Run complete pipeline with DESeq2
pipe.run(
    norm_method="deseq2",
    dea_method="deseq2",
    contrast=["condition", "treated", "control"],
    gene_sets=["KEGG_2021_Human", "GO_Biological_Process_2021"],
)

# Switch to edgeR-like with TMM — one line change
pipe.run(norm_method="tmm", dea_method="edger")

Individual modules

from rnaseq_toolkit import normalize_counts, run_deseq2, plot_volcano
import pandas as pd

counts = pd.read_csv("data/counts.csv", index_col=0)
meta   = pd.read_csv("data/metadata.csv", index_col=0)

# Normalize
norm = normalize_counts(counts, method="deseq2", metadata=meta, design="~condition")

# DEA
results = run_deseq2(counts, meta, "~condition",
                     contrast=["condition", "treated", "control"])

# Visualize
plot_volcano(results, output_path="results/volcano.png")

Input Format

Count matrix (counts.csv): genes as rows, samples as columns.

gene_id,Sample_01,Sample_02,Sample_03,Sample_04,Sample_05,Sample_06
ENSG00000000003,1234,1456,1123,2345,2567,2234
ENSG00000000005,456,512,489,234,198,267
...

Metadata (metadata.csv): samples as rows, variables as columns.

sample_id,condition,batch
Sample_01,control,A
Sample_02,control,A
Sample_03,control,B
Sample_04,treated,A
Sample_05,treated,A
Sample_06,treated,B

Normalization Methods

Method	Description	Use case
deseq2	Median-of-ratios (DESeq2)	Default; DEA with PyDESeq2
tmm	Trimmed Mean of M-values (edgeR)	DEA with edgeR-like
cpm	Counts Per Million	Quick exploration
tpm	Transcripts Per Million (needs gene lengths)	Cross-sample comparison
rpkm	RPKM (needs gene lengths)	Legacy compatibility
vst	Variance Stabilizing Transformation	PCA, heatmaps, clustering
rlog	Regularized log (approximate)	Small sample sizes

Snakemake Workflow

# Edit workflow/config.yaml, then run:
snakemake --cores 4 --configfile workflow/config.yaml

Docker

docker build -t rnaseq-toolkit:0.1.0 .

docker run --rm \
    -v $(pwd)/data:/workspace/data \
    -v $(pwd)/results:/workspace/results \
    rnaseq-toolkit:0.1.0 \
    --counts data/counts.csv \
    --metadata data/metadata.csv \
    --design "~condition" \
    --contrast condition treated control \
    --output results/

Benchmark

rnaseq-toolkit was benchmarked against standalone DESeq2 and edgeR-like implementations on three public GEO datasets:

Dataset	Disease / Condition	GEO Accession	Samples
Batten	CLN2 (rare disease)	GSE210143	12
COVID-19	SARS-CoV-2 infection	GSE157103	126
Breast Ca	TNBC vs. luminal	GSE183947	60

Results demonstrate high concordance (Pearson r > 0.95 for log2FC) between rnaseq-toolkit and standalone tools, with significantly reduced setup time.

Citation

If you use rnaseq-toolkit in your research, please cite:

@software{rnaseq_toolkit_2026,
  title   = {rnaseq-toolkit: Streamlined Differential Expression Analysis
             and Pathway Enrichment Visualization from RNA-seq Data},
  version = {0.1.0},
  year    = {2026},
  url     = {https://github.com/rnaseq-toolkit/rnaseq-toolkit},
  license = {MIT}
}

See also CITATION.cff for full citation metadata.

Dependencies

• PyDESeq2 — DESeq2 in Python
• gseapy — GSEA and Enrichr
• statsmodels — NB GLM for edgeR-like
• scikit-learn — PCA
• seaborn / matplotlib — visualization

License

MIT License — see LICENSE.