Ridge regression wrapper for modelGeneExpression — modelGeneExpression_ridge_regression

Internal function used in modelGeneExpression. It runs ridge regression parallelly across signatures and samples as specified by experiment design.

Usage

modelGeneExpression_ridge_regression_wraper(
  mae,
  yname,
  uname,
  xnames,
  groups,
  standardize,
  parallel,
  precalcmodels,
  ...
)

Arguments

mae: MultiAssayExperiment object such as produced by prepareCountsForRegression.
yname: string indicating experiment in mae to use as the expression input.
uname: string indicating experiment in mae to use as the basal expression level.
xnames: character indicating experiments in mae to use as molecular signatures.
groups: factor representation of design matrix.
standardize: logical flag indicating if the molecular signatures should be scaled. Advised to be set to TRUE.
parallel: parallel argument to internally used cv.glmnet function. Advised to be set to FALSE as it might interfere with parallelization used in modelGeneExpression.
precalcmodels: optional list of precomputed 'cv.glmnet' objects for each molecular signature and sample. The elements of this list should be matching the xnames vector. Each of those elements should be a named list holding 'cv.glmnet' objects for each sample. If provided those models will be used instead of running regression from scratch.
...: arguments passed to glmnet::cv.glmnet.

Value

Named list with elements corresponding to signatures specified in xnames. Each of these is a list holding 'cv.glmnet' objects corresponding to each sample.