regionStats: Find Regions of significance in microarray data

regionStatsR Documentation

Find Regions of significance in microarray data

Description

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.

Usage

  ## 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)

Arguments

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.

Value

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.

Author(s)

Mark Robinson

Examples

## 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)

markrobinsonuzh/Repitools documentation built on March 20, 2024, 6:04 a.m.