Component: LimmaNormalizer

LimmaNormalizer

Normalizes two-channel and single-channel microarrays using the Bioconductor Limma package. The inputs are raw green and red channel intensities and their corresponding background estimates. NA values are allowed. The output are normalized log-ratios; NA values may be present. All input matrices must have the same number of columns, but their column names do not need to be equal. Red inputs and several parameters are not used with single-channel data.

Many of the parameters correspond to Limma function parameters. Refer to Limma documentation for details. The component uses the version of Limma that is installed, but the list of valid parameter values in this documentation are obtained from Limma 3.6.9.

Note that for single-channel, no within-array normalization is done. For two-channel between-array normalization is usually not required.

For Agilent arrays, suggested source columns for green, greenBG, red and redBG are gMedianSignal, gBGMedianSignal, rMedianSignal and rBGMedianSignal.

Version	0.6
Bundle	microarray
Categories	Data Import Normalization
Authors	Kristian Ovaska (kristian.ovaska@helsinki.fi), Viljami Aittomaki (viljami.aittomaki@helsinki.fi)
Issue tracker	View/Report issues
Requires	limma (R-bioconductor)
Source files	component.xml LimmaNormalizer.r
Usage	Example with default values LimmaNormalizer( green = LogMatrix, greenBG = LogMatrix, //red = LogMatrix, //redBG = LogMatrix, //groups = SampleGroupTable, arrayMethod = "none", backgroundMethod = "normexp", loessIterations = 4, loessSpan = 0.3, logratioMethod = "loess", normexpMethod = "mle", outColumnPattern = "%s_%s", redMinusGreen = true )

Version

0.6

Bundle

microarray

Categories

Data Import Normalization

Authors

Kristian Ovaska (kristian.ovaska@helsinki.fi), Viljami Aittomaki (viljami.aittomaki@helsinki.fi)

Issue tracker

View/Report issues

Requires

limma (R-bioconductor)

Source files

component.xml LimmaNormalizer.r

Usage

Example with default values

Inputs

Name	Type	Mandatory	Description
green	LogMatrix	Mandatory	Raw (non-normalized) green channel foreground intensities.
greenBG	LogMatrix	Mandatory	Green channel background intensities.
red	LogMatrix	Optional	Raw (non-normalized) red channel foreground intensities. Not used for single-channel data.
redBG	LogMatrix	Optional	Red channel background intensities. Not used for single-channel data.
groups	SampleGroupTable	Optional	If defined, names of output log-ratio columns and the directions of each log-ratio group are determined from here. For each green/red column pair G/R, there should be an entry of type ratio having the columns G and R as members. If the members are in order G,R (green channel first), the log-ratio for this group is G-R; otherwise, it is R-G. If the input is omitted, redMinusGreen parameter determines the direction and outColumnPattern determines the output column names. Not used for single-channel data.

Name

Type

Mandatory

Description

green

LogMatrix

Mandatory

Raw (non-normalized) green channel foreground intensities.

greenBG

LogMatrix

Mandatory

Green channel background intensities.

red

LogMatrix

Optional

Raw (non-normalized) red channel foreground intensities. Not used for single-channel data.

redBG

LogMatrix

Optional

Red channel background intensities. Not used for single-channel data.

groups

SampleGroupTable

Optional

If defined, names of output log-ratio columns and the directions of each log-ratio group are determined from here. For each green/red column pair G/R, there should be an entry of type ratio having the columns G and R as members. If the members are in order G,R (green channel first), the log-ratio for this group is G-R; otherwise, it is R-G. If the input is omitted, redMinusGreen parameter determines the direction and outColumnPattern determines the output column names. Not used for single-channel data.

Name	Type	Description
logratio	LogMatrix	Normalized log-ratios. The i'th column corresponds to the i'th column in the input matrices.

Name

Type

Description

logratio

LogMatrix

Normalized log-ratios. The i'th column corresponds to the i'th column in the input matrices.

Parameters

Name	Type	Default	Description
arrayMethod	string	"none"	Method for normalizing between arrays. Corresponds to the method parameter of Limma normalizeBetweenArrays function. Valid values: none, scale, quantile, Aquantile, Gquantile, Rquantile or cyclicloess (where the Xquantile options are not valid for single-channel).
backgroundMethod	string	"normexp"	Method for background correction. Corresponds to the method parameter of Limma backgroundCorrect function. Valid values: none, subtract, half, minimum, movingmin, edwards, normexp.
loessIterations	int	4	For LOESS normalization, controls the number of iterations. Corresponds to the iterations parameter of Limma normalizeWithinArrays function. Not used for single-channel data.
loessSpan	float	0.3	For LOESS normalization, controls the smoothing of the LOESS curve. Corresponds to the span parameter of Limma normalizeWithinArrays function. Not used for single-channel data.
logratioMethod	string	"loess"	Method for normalizing log-ratios to account for dyes and other factors. Corresponds to the method parameter of Limma normalizeWithinArrays function. Valid values: none, median, loess. Not used for single-channel data.
normexpMethod	string	"mle"	If backgroundMethod = normexp, this is an additional parameter for background estimation. Valid values: saddle, mle, rma (needs Bioconductor affy package), rma75.
outColumnPattern	string	"%s_%s"	If the sample groups are not defined, this gives the pattern for naming log-ratio output columns. If redMinusGreen is true, the first %s is replaced with the red channel name and the second %s with the green one. If redMinusGreen is false, column names are reversed. Not used for single-channel data.
redMinusGreen	boolean	true	If the sample groups are not defined, this defines the direction of log-ratio computation. If true, all log-ratios are red minus green channel; if false, they are green minus red channel. Not used for single-channel data.

Name

Type

Default

Description

arrayMethod

string

"none"

Method for normalizing between arrays. Corresponds to the method parameter of Limma normalizeBetweenArrays function. Valid values: none, scale, quantile, Aquantile, Gquantile, Rquantile or cyclicloess (where the Xquantile options are not valid for single-channel).

backgroundMethod

string

"normexp"

Method for background correction. Corresponds to the method parameter of Limma backgroundCorrect function. Valid values: none, subtract, half, minimum, movingmin, edwards, normexp.

loessIterations

int

For LOESS normalization, controls the number of iterations. Corresponds to the iterations parameter of Limma normalizeWithinArrays function. Not used for single-channel data.

loessSpan

float

0.3

For LOESS normalization, controls the smoothing of the LOESS curve. Corresponds to the span parameter of Limma normalizeWithinArrays function. Not used for single-channel data.

logratioMethod

string

"loess"

Method for normalizing log-ratios to account for dyes and other factors. Corresponds to the method parameter of Limma normalizeWithinArrays function. Valid values: none, median, loess. Not used for single-channel data.

normexpMethod

string

"mle"

If backgroundMethod = normexp, this is an additional parameter for background estimation. Valid values: saddle, mle, rma (needs Bioconductor affy package), rma75.

outColumnPattern

string

"%s_%s"

If the sample groups are not defined, this gives the pattern for naming log-ratio output columns. If redMinusGreen is true, the first %s is replaced with the red channel name and the second %s with the green one. If redMinusGreen is false, column names are reversed. Not used for single-channel data.

redMinusGreen

boolean

true

If the sample groups are not defined, this defines the direction of log-ratio computation. If true, all log-ratios are red minus green channel; if false, they are green minus red channel. Not used for single-channel data.

Test cases

Test case	Parameters▼	IN green	IN greenBG	IN red	IN redBG	IN groups	OUT logratio
case1_nonorm	properties	green	greenBG	red	redBG	groups	logratio
	backgroundMethod=none, logratioMethod=none
case2_subtract	properties	green	greenBG	red	redBG	groups	logratio
	backgroundMethod=subtract, logratioMethod=none
case3_subtract2	properties	green	greenBG	red	redBG	(missing)	logratio
	backgroundMethod=subtract, logratioMethod=none, redMinusGreen=false
case4_defaults	(missing)	green	greenBG	red	redBG	(missing)	logratio

Test case

Parameters▼

IN
green

IN
greenBG

IN
red

IN
redBG

IN
groups

OUT
logratio

case1_nonorm

properties

green

greenBG