seurust

Rust/extendr backend for Seurat's performance-critical native routines. Install alongside Seurat to use the same function signatures with a Rust backend.

Requirements

• R (>= 4.0.0)
• Rust toolchain: rustc and Cargo (>= 1.81)
• On Windows: Rtools plus Rust

No C++ compiler or RcppEigen headers are needed; all compiled code is Rust.

Install

From CRAN (after acceptance)

install.packages("seurust")

From r-universe

install.packages(
  "seurust",
  repos = c("https://NebilI.r-universe.dev", "https://cloud.r-project.org")
)

From GitHub

if (!requireNamespace("remotes", quietly = TRUE)) {
  install.packages("remotes")
}

remotes::install_github("NebilI/seurust", subdir = "seurust")

From a release tarball

Download seurust_*.tar.gz from GitHub Releases, then:

install.packages("path/to/seurust_0.1.5.tar.gz", repos = NULL, type = "source")

Example

library(Seurat)
library(seurust)
library(Matrix)

mat <- Matrix::sparseMatrix(i = c(1, 3, 2), j = c(1, 2, 3), x = 1:3, dims = c(3, 3))
all.equal(
  Seurat:::LogNorm(mat, 1e4, FALSE),
  seurust::LogNorm(mat, 1e4, FALSE)
)

Differences from Seurat's C++ entry points

Every exported function takes the same arguments and returns the same kind of object as the Seurat function of the same name. Integer inputs are accepted wherever Seurat's Rcpp signature would coerce them (for example integer neighbor ids in fast_dist or an integer nn_ranked matrix in ComputeSNN).

One deliberate difference: Seurat's RowMergeMatrices only accepts row-compressed (RsparseMatrix) inputs and errors on a dgCMatrix. seurust::RowMergeMatrices accepts either layout and converts first; for RsparseMatrix inputs the result is identical.

Further documentation

Development, packaging, and contribution notes live in the GitHub repository.