rsfgseapy 0.3.4

Fast Gene Set Enrichment Analysis in Rust (Python bindings)

rsfgseapy

Python bindings for rsfgsea, a Rust implementation of preranked fgsea-compatible gene set enrichment analysis.

What It Exposes

The package currently exposes one public entrypoint:

• run_gsea_py(...)

The API intentionally keeps fgsea-style parameter names and execution modes:

• mode="fgsea"
• mode="simple"
• mode="multilevel"
• nPermSimple
• seed
• nperm
• minSize
• maxSize
• sampleSize
• scoreType
• gseaParam

Installation

From PyPI:

pip install rsfgseapy

From a repository:

git clone https://github.com/deminden/rsfgsea
cd rsfgsea
cd crates/rsfgseapy
maturin develop --release

Input Shape

ranks

• Python mapping of gene -> score
• values must be finite numeric scores

gmt_path

• path to a GMT file

Performance Snapshot

The Python package calls the same Rust backend as the CLI and R wrapper. Current local benchmark snapshots from the main repository, measured on an AMD Ryzen 7950X3D:

• representative Criterion benchmark, simple: 2.282 s for 10k genes, 1k pathways, 10k permutations
• representative Criterion benchmark, multilevel: 3.438 s for 10k genes, 1k pathways, nPermSimple=1000
• file-backed comparison, multilevel large workload, 16 workers: Rust 105 ms vs R 977 ms (9.3x faster)
• file-backed comparison, simple large workload, 16 workers: Rust 674 ms vs R 798 ms (1.18x faster)
• real muscle-comparison validation workload: Rust 81 MB peak RSS vs R 329 MB peak RSS (4.1x lower)

Full benchmark setup, thread-scaling tables, and parity notes are in:

• https://github.com/deminden/rsfgsea/blob/main/docs/reproducibility.md

Minimal Example

For most users, wrapper mode with defaults is the right starting point.

import rsfgseapy

results = rsfgseapy.run_gsea_py(
    ranks={"GENE_A": 2.0, "GENE_B": 1.0, "GENE_C": -1.0, "GENE_D": -2.0},
    gmt_path="pathways.gmt",
)

for row in results:
    print(row["pathway"], row["pval"])

Full Example

import rsfgseapy

ranks = {
    "GENE_A": 3.2,
    "GENE_B": 1.7,
    "GENE_C": -2.4,
    "GENE_D": -3.1,
}

results = rsfgseapy.run_gsea_py(
    ranks=ranks,
    gmt_path="pathways.gmt",
    mode="fgsea",
    gpu=False,
    nPermSimple=1000,
    seed=None,
    nperm=None,
    minSize=1,
    maxSize=None,
    eps=1e-50,
    sampleSize=101,
    scoreType="std",
    gseaParam=1.0,
    nproc=0,
)

for row in results:
    print(row["pathway"], row["nes"], row["pval"])

Plotting

import rsfgseapy

rsfgseapy.write_enrichment_plot_png_py(
    ranks={"GENE_A": 2.0, "GENE_B": 1.0, "GENE_C": -1.0, "GENE_D": -2.0},
    pathway_genes=["GENE_A", "GENE_B"],
    output_path="enrichment.png",
    pathway_name="PW_A",
    dpi=300,
    title="PW_A",
)

For multi-pathway summaries:

import rsfgseapy

rsfgseapy.write_gsea_table_plot_png_py(
    ranks={"GENE_A": 2.0, "GENE_B": 1.0, "GENE_C": -1.0, "GENE_D": -2.0},
    pathways=[("PW_A", ["GENE_A", "GENE_B"]), ("PW_B", ["GENE_C", "GENE_D"])],
    results=[
        {"pathway": "PW_A", "nes": 1.5, "pval": 0.01, "padj": 0.02},
        {"pathway": "PW_B", "nes": -1.4, "pval": 0.03, "padj": 0.05},
    ],
    output_path="table.png",
    dpi=300,
)

All plotting parameters are available in the Python API; the examples above keep only the most common publication-oriented overrides visible.

For the full cross-interface plotting guide, see:

• https://github.com/deminden/rsfgsea/blob/main/docs/plotting.md

nPermSimple vs nperm

These two names come from fgsea and they are not interchangeable.

nPermSimple

• the normal simple-stage permutation count
• used by default in wrapper mode
• tune this when you want a different wrapper screening budget

nperm

• explicit fixed-permutation override
• in wrapper mode, setting nperm forces simple-mode execution instead of multilevel refinement
• leave this as None unless you intentionally want simple mode

Practical rule:

• leave seed=None for a fresh random run, or set seed=<int> for reproducibility
• light users: keep nperm=None
• use nPermSimple to tune the default wrapper behavior
• only set nperm when you deliberately want fixed-permutation simple execution

Returned Results

Each result row is a dictionary with:

• pathway
• size
• es
• nes
• pval
• padj
• log2err
• leading_edge

leading_edge is returned as a Python list of genes.

GPU Support

gpu=True enables the hybrid GPU path when the extension is built with the gpu feature.

Current behavior:

• GPU accelerates simple-stage screening
• CPU performs parity-focused multilevel refinement

If the extension is built without GPU support, gpu=True raises a runtime error.

Supported Python Versions

The package metadata currently targets Python 3.8 and newer.

Project Links

• Repository: https://github.com/deminden/rsfgsea
• Main project docs: https://github.com/deminden/rsfgsea/tree/main/docs
• Rust crate: https://crates.io/crates/rsfgsea