FastMNN batch correction in Rust — rs

This function implements the (fast) MNN algorithm from Haghverdi, et al. Instead of working on the full matrix, it uses under the hood PCA and generates an aligned embedding space.

Usage

rs_mnn(
  f_path_gene,
  cell_indices,
  gene_indices,
  batch_indices,
  precomputed_pca,
  mnn_params,
  verbose,
  seed
)

Arguments

f_path_gene: String. Path to the counts_genes.bin file.
cell_indices: Integer. The cell indices to use. (0-indexed!)
gene_indices: Integer. The gene indices to use. (0-indexed!) Ideally these are batch-aware highly variable genes.
batch_indices: Integer vector. These represent to which batch a given cell belongs.
precomputed_pca: Optional PCA matrix. If you want to provide a pre-computed matrix.
mnn_params: List. Contains all of the fastMNN parameters.
verbose: Boolean. Controls verbosity of the function.
seed: Integer. Seed for reproducibility purposes.

Value

The batch-corrected embedding space.