optimizeNewLambda {rliger}

Perform factorization for new lambda value

Description

Uses an efficient strategy for updating that takes advantage of the information in the existing factorization; always uses previous k. Recommended mainly when re-optimizing for higher lambda and when new lambda value is significantly different; otherwise may not return optimal results.

Usage

optimizeNewLambda(
  object,
  lambdaNew,
  nIteration = 30,
  seed = 1,
  verbose = getOption("ligerVerbose"),
  new.lambda = lambdaNew,
  max.iters = nIteration,
  rand.seed = seed,
  thresh = NULL
)

Arguments

object

liger object. Should have integrative factorization (e.g. runINMF) performed in advance.

lambdaNew

Numeric regularization parameter. Larger values penalize dataset-specific effects more strongly.

nIteration

Number of block coordinate descent iterations to perform. Default 30.

seed

Random seed to allow reproducible results. Default 1. Used by runINMF factorization.

verbose

Logical. Whether to show information of the progress. Default getOption("ligerVerbose") which is TRUE if users have not set.

new.lambda, max.iters, rand.seed

These arguments are now replaced by others and will be removed in the future. Please see usage for replacement.

thresh

Deprecated. New implementation of iNMF does not require a threshold for convergence detection. Setting a large enough nIteration will bring it to convergence.

Value

Input object with optimized factorization values updated. including the W matrix in liger object, and H and V matrices in each ligerDataset object in the datasets slot.

See Also

runINMF, optimizeNewK, optimizeNewData

Examples

pbmc <- normalize(pbmc)
pbmc <- selectGenes(pbmc)
pbmc <- scaleNotCenter(pbmc)
if (requireNamespace("RcppPlanc", quietly = TRUE)) {
    # Only running a few iterations for fast examples
    pbmc <- runINMF(pbmc, k = 20, nIteration = 2)
    pbmc <- optimizeNewLambda(pbmc, lambdaNew = 5.5, nIteration = 2)
}