regionStats: Find Regions of significance in microarray data
In Repitools: Epigenomic tools

Description Usage Arguments Value Author(s) Examples

The function finds the highest smoothed score cutoff for a pre-specified FDR. Smoothing is performed over a specified number of basepairs, and regions must have a minimum number of qualifying probes to be considered significant. The FDR is calculated as the ratio of the number of significant regions found in a permutation-based test, to the number found in the actual experimental microarray data.

  ## S4 method for signature 'matrix'
regionStats(x, design = NULL, maxFDR=0.05, n.perm=5, window=600, mean.trim=.1, min.probes=10, max.gap=500, two.sides=TRUE, ndf, return.tm = FALSE, verbose=TRUE)
  ## S4 method for signature 'AffymetrixCelSet'
regionStats(x, design = NULL, maxFDR=0.05, n.perm=5, window=600, mean.trim=.1, min.probes=10, max.gap=500, two.sides=TRUE, ind=NULL, return.tm = FALSE, verbose=TRUE)

`x`	An `AffymetrixCelSet` or `matrix` of array data to use.
`design`	A design matrix of how to manipulate
`maxFDR`	Cutoff of the maximum acceptable FDR
`n.perm`	Number of permutations to use
`window`	Size of window, in base pairs, to check for
`mean.trim`	A number representing the top and bottom fraction of ordered values in a window to be removed, before the window mean is calculated.
`min.probes`	Minimum number of probes in a window, for the region to qualify as a region of significance.
`max.gap`	Maximum gap between significant probes allowable.
`two.sides`	Look for both significant positive and negative regions.
`ind`	A vector of the positions of the probes on the array
`ndf`	The Nimblegen Definition File for Nimblegen array data.
`return.tm`	If TRUE, the values of the trimmed means of the intensities and permuted intensities are also retuned from the function.
`verbose`	Whether to print the progress of processing.

A RegionStats object (list) with elements

`regions`	A list of `data.frame`. Each `data.frame` has columns `chr`, `start`, `end`, `score`.
`tMeanReal`	Matrix of smoothed scores of intensity data. Each column is an experimental design.
`tMeanPerms`	Matrix of smoothed scores of permuted intensity data. Each column is an experimental design.
`fdrTables`	List of table of FDR at different score cutoffs. Each list element is for a different experimental design.

Mark Robinson

## Not run: 
library(Repitools)
library(aroma.affymetrix)

# assumes appropriate files are at annotationData/chipTypes/Hs_PromPR_v02/
cdf <- AffymetrixCdfFile$byChipType("Hs_PromPR_v02",verbose=-20)
cdfU <- getUniqueCdf(cdf,verbose=-20)

# assumes appropriate files are at rawData/experiment/Hs_PromPR_v02/
cs <- AffymetrixCelSet$byName("experiment",cdf=cdf,verbose=-20)
mn <- MatNormalization(cs)
csMN <- process(mn,verbose=-50)
csMNU <- convertToUnique(csMN,verbose=-20)

#> getNames(cs)
# [1] "samp1"  "samp2"  "samp3"  "samp4"

design <- matrix( c(1,-1,rep(0,length(cs)-2)), ncol=1, dimnames=list(getNames(cs),"elut5_L-P") )

# just get indices of chr7 here
ind <- getCellIndices(cdfU, unit = indexOf(cdfU, "chr7F"), unlist = TRUE, useNames = FALSE)

regs <- regionStats(csMNU, design, ind = ind, window = 500, verbose = TRUE)

## End(Not run)