Skip to contents

Run multiple NMF (HALS) restarts on MetaCells

rs_nmf_multi_mc.Rd

[Experimental] Assumes that the sparse data is pre-filtered for the cells/genes you wish to include. Indices in the sparse data need to be 0-indexed. Both data layers hold the supplied values, so which assay NMF runs on is decided by what is passed in, not by use_second_layer.

Usage

rs_nmf_multi_mc(
  sparse_data,
  k,
  preprocessing,
  use_second_layer,
  nmf_hals_params,
  n_runs,
  seed,
  verbose
)

Arguments

sparse_data

A named list with data, indptr, indices, nrow, ncol and cs_type. Shape is (metacells, genes).

k

Integer. Number of latent factors per run.

preprocessing

String. One of c("none", "sd", "sqrt_sd").

use_second_layer

Boolean. Shall the second data layer be used.

nmf_hals_params

Named list. Contains the NMF parameters, see params_nmf_hals(). The nmf_init field is ignored, restarts always use random initialisation.

n_runs

Integer. Number of random restarts.

seed

Integer. Base random seed. Run i uses seed + i.

verbose

Integer. 0L - quiet; 1L - normal verbosity; 2L - detailed verbosity.

Value

A list with the following items

  • w_all - Column-bound W matrices across all runs, shape n_meta_cells x (k * n_runs).

  • h_per_run - List of H matrices, each k x n_genes.

  • losses - Numeric vector. Final reconstruction loss per run.

  • converged - Logical vector. Convergence flag per run.

  • best_idx - Integer. 1-indexed position of the run with the lowest final loss.