rapids-singlecell-complete

What problem does it solve?

RAPIDS Single-Cell Complete provides GPU-accelerated tooling and documentation to speed up large-scale single-cell RNA-seq analyses on AnnData objects, enabling faster preprocessing, dimensionality reduction, clustering, and downstream interpretation.

Core Features & Use Cases

• GPU-accelerated preprocessing: QC metrics, normalization, HVG selection, and scaling on GPU with RAPIDS.
• Dimensionality reduction & clustering: PCA, UMAP, t-SNE, Louvain/Leiden with GPU acceleration for large datasets.
• Advanced workflows: Multi-GPU workflows with Dask-CUDA, out-of-core processing, and spatial analysis (ligand-receptor interactions) with RAPIDS ecosystem integration.
• Use Case: Process a 1-2 million cell scRNA-seq dataset by moving data to the GPU, performing batch correction, and generating embeddings and clusters within minutes rather than hours.

Quick Start

Install RAPIDS and this skill, load your AnnData object, move data to the GPU, and run the standard GPU-accelerated workflow: QC metrics, HVG detection, normalization, log1p, HVG filtering, PCA, neighbors, and UMAP/TSNE, followed by clustering and visualization.