AALoess: Normalises a set of gene expression data files using LOESS

In agilp: Agilent expression array processing package

Description

This script normalises a set of gene expression data files against a predefined reference data set. The normalisation uses LOESS regression. The major assumption is that the actual data and the reference set have a similar epxression frequency distribution. Typically, the reference set used the is average of a large number of datasets such as the output of Baseline script in this package. The script also calculates the Sum of Squared Errror (SSE) between each data point of the normalised data set and the reference, and plots the results as a histogram of the SSE of all the data sets processed. This function is useful to identify outliers, which are usually due to problems in sample prepration/processing, which are frequently not identified by using the internal scanner quality performance indicators.

Usage

AALoess(input=file.path(system.file(package="agilp"),"input",""),output=file.path(system.file(package="agilp"),"output",""), baseline=file.path(system.file(package="agilp"),"input1","testbase.txt"),LOG="TRUE")

Arguments

input	full path of directory where input data files are put; default is a folder named input within the agilp package directory; the input directory must contain ONLY the set of files to be processed. The input files must contain two columns only, the source identifier in the first column, and the expression data in the second column; see example files
output	full path of directory where output data files are put; default is a folder named output within the agilp package directory
LOG	if NORM="LOG", the data are log base 2 transformed before normalisation. The default is NORM="LOG"
baseline	full path of file containing the reference data set for normalisation; the default is agilp/input1/baseline.txt

Value

normalised datafiles	One new file is written to the output folder for each input file. The format of the output files is the same as the input, but the data has been normalised. The normalised files are identified by the prefix nr_ or ng_
SSE_date.txt file	A file in the output folder containing a list of all the filenames in the input folder, and the corresponding SSE value

.

Author(s)

Benny Chain; [e-mail]

References

In preparation; for further detail on normalisation see also Chain B, Bowen H, Hammond J, Posch W, Rasaiyaah J, Tsang J, Noursadeghi M. Error, reproducibility and sensitivity: a pipeline for data processing of Agilent oligonucleotide expression arrays. BMC Bioinformatics. 2010 Jun 24;11:344.

See Also

AAProcess filenamex Loader Baseline IDswop Equaliser

Examples

#Takes four files of raw data (output of AAProcess, in dataset/raw folder) , LOess normalises them and saves them in output folder
inputdir<-file.path(system.file(package="agilp"),"extdata","raw/","", fsep = .Platform$file.sep)
outputdir<-file.path(system.file(package="agilp"),"output/", "", fsep = .Platform$file.sep)
baselinedir<-file.path(system.file(package="agilp"),"extdata","testbase.txt", fsep = .Platform$file.sep)
AALoess(input=inputdir, output=outputdir, baseline = baselinedir, LOG="TRUE") 
## Not run: 
#to remove these files again and empty the output directory use 
unlink(paste(file.path(system.file(package="agilp"),"output",""),"*.*",sep=""), recursive=FALSE)

## End(Not run)
	

agilp documentation built on Nov. 8, 2020, 5:45 p.m.