Skip to contents

Maximum Margin Clustering (MMC) Transformation (Memory-Optimized)

Source: R/scLearn_model_learning.R
runMMC.Rd

Performs MMC metric learning for single-cell data with reduced memory usage. Incorporates low-rank approximation and other optimizations.

Usage

runMMC(
  high_varGenes,
  expression_profile,
  sample_information,
  C = 1,
  max_iter = 50,
  seed = 1,
  verbose = TRUE,
  rank = NULL,
  batch_size = NULL
)

Arguments

high_varGenes

Character vector of high-variance genes

expression_profile

Numeric matrix (genes x cells)

sample_information

Named vector of cell type labels

C

Trade-off parameter (default: 1.0)

max_iter

Maximum iterations (default: 50)

seed

Random seed (default: 1)

verbose

Print progress messages (default: TRUE)

rank

Optional rank for low-rank approximation (default: NULL)

batch_size

Number of cells to process at once if using batching (default: NULL)

Value

List containing:

  • expression_profile_trans: Transformed matrix (genes x cells)

  • expression_profile_origin: Original matrix

  • trans_matrix: Learned transformation matrix

  • sample_information: Input cell labels