selectGene-methods: Gene selection using SPLS model

In sunny-zq/INGRID: InGRiD: Semi-supervised Identification of Cancer Subgroups using Survival Outcomes and Overlapping Grouping Information

Description

Sparse Partial Least Square (SPLS) step for gene selection and dimension reduction.By applying SPLS to each pathway, we achieve the goal of gene selection and dimension reduction at the same time.

Usage

## S4 method for signature 'Prefiltered'
selectGene(object, K = 5, etas = seq(0.1, 0.9,
  0.1), fold = 5, se1 = TRUE, method = "plik", par = FALSE,
  foldid = NULL, seed = 123)

Arguments

object	output list of prefilter step
K	the maximum number of hidden features in spls.
etas	Thresholding parameter. eta should be between 0 and 1.
fold	The number of folds to use to perform the cross-validation process.
se1	one standard error criteria, default = TRUE
method	if auc is used, this is the maximum auc criteria, if partial likelihood(plik) is used, this is the min deviance criteria. Default = TRUE
par	parallel option for running the code, default = FALSE
foldid	the number of folds to use to perform the cross-validation process.
seed	random seed that was set, default = 123.

Examples

data(TCGA)
geneRegroup.result=geneRegroup(plist=TCGA$pathList )
prefilter.results=prefilter( data=TCGA$geneexpr, time=TCGA$t, status=TCGA$d,
                             plist=geneRegroup.result@gset )
gene.results=selectGene( object=prefilter.results, fold=5, K=5, etas=c(0.1,0.9),seed=123)

sunny-zq/INGRID documentation built on Oct. 15, 2019, 6:45 p.m.