Return the coefficients from the ARRm linear model

Share:

Description

For each probe type, it returns the coefficients of the linear model used in the ARRm normalization. Since the model is applied to each percentile separately, different coefficients are returned for every percentile. Residuals are returned as well.

Usage

1
getCoefficients(quantiles,designInfo,backgroundInfo,outliers.perc=0.02)

Arguments

quantiles

A list containing three matrices. "$green", "$red" and "$II" must contain respectively the matrices of percentiles obtained from a "betaMatrix" for the Type I Green probes, Type I Red probes and Type II probes. See getQuantiles.

designInfo

matrix returned by getDesignInfo

backgroundInfo

matrix returned by getBackground

outliers.perc

Percentage of outliers to be removed in the regression. By default, set to 0.02

Value

Returns a list containing three lists of coefficients for each probe type. ($green to access coefficients for Type I green probes, $red to access coefficients for Type I red probes and $II to access coefficients for Type II probes). Each list of coefficients contains five subfields. res is a matrix of residuals for the linear model across percentiles (a vector of residuals for each percentile), background.vector is a vector containing the regression coefficients for background intensity across percentiles; dyebias.vector is a vector containing the regression coefficients for dye bias across percentiles; chip.variations is a matrix of chip variations estimated by the linear model; rows correspond to percentiles, columns correspond to chips; position.variations is a matrix of position deviation from the chip mean estimated by the linear model; rows correspond to percentiles, columns correspond to positions.

Author(s)

Jean-Philippe Fortin <jfortin@jhsph.edu>

Examples

1
2
3
4
5
6
7
8
data(greenControlMatrix)
data(redControlMatrix)
data(sampleNames)
data(betaMatrix)
backgroundInfo=getBackground(greenControlMatrix,redControlMatrix)
designInfo=getDesignInfo(sampleNames)
quantiles=getQuantiles(betaMatrix)
coefficients=getCoefficients(quantiles,designInfo,backgroundInfo)