Grid-search to preserve small clusters (k.param & resolution)

Grid-search to preserve small clusters (k.param & resolution)

Usage

tune_clustering(
  obj,
  dims_use = 1:30,
  k_vals = c(10, 12, 15, 20),
  res_vals = seq(0.2, 1.6, by = 0.2),
  algorithm = 4,
  prune = 0.1,
  small_frac_target = c(0.005, 0.05),
  seed = 1
)

Arguments

obj

Seurat object

dims_use

Integer vector of PC indices

k_vals

Integer vector of k.param candidates

res_vals

Numeric vector of resolution candidates

algorithm

Integer; clustering algorithm (e.g., Leiden=4)

prune

Numeric; prune.SNN

small_frac_target

Length-2 numeric; desired min cluster fraction range

seed

Integer random seed

Value

List with components $stats (grid) and $pick (chosen row)