We kept running into the same problem: pipelines are in Python, but the stats we needed (edgeR, limma-voom, clusterProfiler) are in R. Writing rpy2 boilerplate for every analysis got old fast, so we built a wrapper library.

pip install rosetta-bioc

edgeR:

from rosetta import edger
results = edger(counts=counts_df, metadata=meta_df, design="~ condition")

limma-voom:

from rosetta import limma_voom
results = limma_voom(counts=counts_df, metadata=meta_df, design="~ batch + condition")

GO enrichment:

from rosetta import enrich_go
go = enrich_go(gene_list=sig_genes, org_db="org.Hs.eg.db")

These call the real R packages (not reimplementations) - results match what you'd get in an R console. Supports contrasts, LFC thresholds, shrinkage, multi-factor designs.

Requires R 4.0+ with the Bioconductor packages installed. MIT licensed. GSoC 2026.

Code: https://github.com/rosetta-bioc/rosetta

Happy to hear feedback or feature requests.