diffGeneAnalysis (version 1.54.0)

normalize: Normalization of microarray data

Description

Normalization of data utilizing information obtained from background fluoresence.Background fluoresce intensity values are used to determine a Gaussian distribution of lowly expressed genes,yielding the background estimates(mean and standard deviation).

Usage

normalize(rawdata, numSlides,ctrl,expm,ctrlbg,expmbg)

Arguments

rawdata

rawdata is matrix of microarray data.The first column consists of gene Names and the first row consists of headers.

numSlides

numSlides represents the total number of chips/slides in the microarray dataset including control and experiment.Control slides are always followed by experiment slides from left to right in the matrix.

ctrl

ctrl represents the total number of control chips in the microarray dataset.

expm

expm represents the total number of experiment chips in the microarray dataset.

ctrlbg

ctrlbg represents the percent of data to pick for background computation of the control chips.30 percent is the default.

expmbg

expmbg represents the percent of data to pick for background computation of the experiment chips.30 percent is the default.

Value

A matrix of normalized values of rawdata

Details

The normalization algorithm trims the data based on innitial emperical estimates of the mean and standard deviation.All data beyond +/-2SD of the mean are cut iteratively.This procedure is repeated until no more cuts can be made.The trimmed data is then subjected to a non linear curve fitting procedure. The user is presented with six different pictures obtained using bars 2,3,4,4.5,5,and5.5 as mean. The user is given the freedom to select the best visual estimate of background. The user selected parameters are used to perform a z-Trnasformation on the data.The percent of data selected to compute background depends on the data obtained.The default is 30 percent.A normal distributed histogram should confirm that, else the user is allowed to pick a percent and make changes until the user sees a normal distributed histogram.Upon running normalize the user is presented with a set of 6 histograms. If the user is not happy with the default 30 percent, the user should go ahead and select a mean and confirm curvefit,then select 'no' to confirm histogram distribution.The user will be presented with a new set of 6 histograms. This process is repeated until the user selects the best Histogram distribution.This process is repeated for each individual chip.

References

Dozmorov I,Centola,M. An associative analysis of gene expression array data. Bioinformatics.2003 Jan22;19(2):204-11

Knowlton N,Dozmorov I, Centola M. Microarray data Analysis Tool box(MDAT): for normalization,adjustment and analysis of gene expression data. Bioonformatics.2004 Dec 12;20(18):3687-90

Examples

#rawdata is loaded in the package. Run example as follows:
#Read the description file for best results.
#data(rawdata)
#normalize(rawdata,7,3,4,0.15,0.60)