optimizeNewK: Perform factorization for new value of k

In rliger: Linked Inference of Genomic Experimental Relationships

Perform factorization for new value of k

Description

This uses an efficient strategy for updating that takes advantage of the information in the existing factorization. It is most recommended for values of k smaller than current value, where it is more likely to speed up the factorization.

Usage

optimizeNewK(
  object,
  k.new,
  lambda = NULL,
  thresh = 1e-04,
  max.iters = 100,
  rand.seed = 1,
  verbose = TRUE
)

Arguments

object	liger object. Should call optimizeALS before calling.
k.new	Inner dimension of factorization (number of factors)
lambda	Regularization parameter. By default, this will use the lambda last used with optimizeALS.
thresh	Convergence threshold. Convergence occurs when |obj0-obj|/(mean(obj0,obj)) < thresh (default 1e-4).
max.iters	Maximum number of block coordinate descent iterations to perform (default 100).
rand.seed	Random seed to set. Only relevant if k.new > k. (default 1)
verbose	Print progress bar/messages (TRUE by default)

Value

liger object with H, W, and V slots reset.

Examples

ligerex <- createLiger(list(ctrl = ctrl, stim = stim))
ligerex <- normalize(ligerex)
ligerex <- selectGenes(ligerex)
ligerex <- scaleNotCenter(ligerex)
k <- 5
# Minimum specification for fast example pass
ligerex <- optimizeALS(ligerex, k = k, max.iters = 1)
if (k != 5) {
    ligerex <- optimizeNewK(ligerex, k.new = k, max.iters = 1)
}

rliger documentation built on Nov. 9, 2023, 1:07 a.m.