deg.jsGPU-Accelerated Differential Expression Analysis

Wilcoxon rank-sum test for single-cell RNA-seq data, running entirely in the browser

Get Started

API Reference

⚡

GPU Acceleration

WebGPU (default) with WebGL2 fallback. 10-100x faster than CPU implementations.

🧬

Scanpy Compatible

TypeScript implementation of sc.tl.rank_genes_groups (Wilcoxon method). Same statistical foundation.

📊

Large Dataset Support

Process 3M+ cells with streaming mode. Automatic chunked processing for memory efficiency.

🌐

Browser-Native

No server required. Run differential expression analysis entirely in the browser.

Quick Example

typescript

import { rankGenesGroups } from 'deg.js';

const result = await rankGenesGroups(expression, groups, {
  nGenes: 100,
  corrMethod: 'benjamini-hochberg',
});

for (const [groupName, data] of result.groups) {
  console.log(`Top genes for ${groupName}:`, data.names.slice(0, 10));
}

Performance Comparison

Tested on Intel i7-1265U, 32GB RAM, Intel Iris Xe Graphics (integrated GPU)

Dataset Size	Python Scanpy	deg.js
10K cells	~5s	~0.5s
100K cells	~30s	~3s
1M cells	~5min	~60s

First Run Initialization

The first analysis includes WebGPU/WebGL2 initialization (shader compilation, pipeline creation). Subsequent runs are significantly faster as the GPU context is reused.

Browser Support

Browser	WebGPU	WebGL2
Chrome 113+	✅	✅
Edge 113+	✅	✅
Firefox 141+	✅	✅
Safari 17+	✅	✅