SimBindProfiles (version 1.10.0)

findBoundCutoff: Find the bound.cutoff

Description

Define the bound.cutoff using either the normalNull or the twoGaussiansNull method.

Usage

findBoundCutoff(xSet, method = c("normalNull", "twoGaussiansNull"), mean.method = "mode",  pvalue = FALSE, fdr = FALSE, pvalPlot = FALSE)

Arguments

xSet

object of class ExpressionSet

method

a character string equal to “normalNull” or “twoGaussiansNull”

mean.method

a characyer string equal to “mode” or “zero”. This is used in combination with the “normalNull” method.

pvalue

decimal specifying the p-value cutoff (either pvalue or fdr can be used not both!)

fdr

decimal specifying the fdr cutoff (either pvalue or fdr can be used not both!)

pvalPlot

logical, if TRUE the pvalue histogram is written

Value

Details

We implemented two methods to set the bound.cutoff, probes above this threshold are considered “bound”. The twoGaussiansNull method established in the Ringo package (Toedling et al., 2007), by which the data is assumed to follow a mixture of two Gaussian distributions. The one Gaussian with the lower mean value is assumed to be the null distribution and probe levels are assigned p-values based on this null distribution. Alternatively the user can select the normalNull method instead which assumes the null distribution is normal and symmetrical around the mode (or zero). For both methods the user can decide if the resulting p-values are to be adjusted for multiple testing (fdr) or selected by a p-value threshold. The function also provides QC plots for the twoGaussiansNull and an optional p-value histogram.

References

Toedling J., Skylar O., Krueger T, Fischer J.J., Sperling S., Huber W. 2007 Ringo - an R/Bioconductor package for analyzing ChIP-chip readouts. BMC Bioinformatics, 8:221

Examples

  dataPath <- system.file("data",package="SimBindProfiles")
  load(file.path(dataPath,"SGR.RData"))
  bound.cutoff <- findBoundCutoff(SGR, method="twoGaussiansNull", fdr=0.25)