metalsinglecell#
GPU-accelerated single-cell analysis on Apple Silicon.
A Metal/MLX re-implementation of rapids-singlecell — drop-in replacements for the core scanpy and squidpy functions that run on the M-series GPU. Swap the import prefix and your existing pipeline runs on the GPU.
import scanpy as sc
import metalsinglecell as msc
adata = sc.datasets.pbmc3k()
msc.pp.normalize_total(adata, target_sum=1e4)
msc.pp.log1p(adata)
msc.pp.highly_variable_genes(adata, n_top_genes=2000)
msc.pp.pca(adata)
msc.pp.neighbors(adata)
msc.tl.leiden(adata, backend="gpu")
msc.tl.umap(adata)
sc.pl.umap(adata, color="leiden")
Installation
Get started — pip install metalsinglecell and environment setup.
Tutorials
Executable notebooks mirroring the rapids-singlecell workflows.
API reference
Every pp / tl / gr function, documented.
GitHub
Source, issues, and releases.