Prune a similarity matrix (TOM or adjacency) to enhance sparsity — prune_matrix • QUESTIONS

Supports thresholding or top-k nearest neighbors to reduce matrix density, improving interpretability and biological signal in diffusion tasks.

Usage

prune_matrix(
  mat,
  value = NULL,
  auto = FALSE,
  target_sparsity = 0.05,
  verbose = TRUE
)

Arguments

mat: A square similarity matrix (e.g., TOM or adjacency)
value: A numeric threshold (if method = "threshold") or number of neighbors (if method = "topk")
auto: If TRUE, determine threshold automatically to achieve target sparsity
target_sparsity: Desired sparsity level (only used if auto = TRUE and method = "threshold")
verbose: If TRUE, prints pruning diagnostics

Value

A pruned matrix with zeroed weak connections