Package 'TurboNorm'

A fast scatterplot smoother with applications for microarray normalization

Description

A fast scatterplot smoother based on B-splines with second order difference penalty. Functions for microarray normalization of single-colour data i.e. Affymetrix/Illumina and two-colour data supplied as marray MarrayRaw-objects or limma RGList-objects are available.

Details

Package:	TurboNorm
Type:	Package
Version:	1.7.2
Date:	2013-29-01
License:	LGPL
LazyLoad:	yes

This package contains an implementation of piecewise constant P-splines of Eilers and Marx (1996) that can be used for normalization of either single- or two-colour data. For two-colour data objects of type RGList from the limma package and MarrayRaw from the package marray can be normalized using the function pspline(). For single colour microarray data wrapper functions are written based on the affy package functions normalize.loess() and normalize.AffyBatch.loess() namely normalize.pspline() and normalize.AffyBatch.pspline(). Also a panel.pspline() is available for adding the smoothed curve to lattice graphics panels.

Note

The package pspline (S original by Jim Ramsey, R port by Brian Ripley) implements the B-spline/Natural Cubic Spline smoother

Author(s)

Chantal van Leeuwen and Maarten van Iterson Maintainer: Maarten van Iterson<[email protected]>

References

van Iterson M, Duijkers FA, Meijerink JP, Admiraal P, van Ommen GJ, Boer JM, van Noesel MM, Menezes RX (2012). A novel and fast normalization method for high-density arrays. SAGMB, 11(4).

Paul .H.C. Eilers and Brain D. Marx (1996). Flexible smoothing with B-splines and Penalties. Statistical Science, Vol 11, No. 2, 89-121.

See Also

turbotrend, pspline, normalize.pspline, normalize.AffyBatch.pspline, panel.pspline

---

CpG island DNA methylation array data

Description

CpG island DNA methylation array data of a neuro-ectodermal cell line that was treated with a demethylating agent

Usage

data(methylation)

Format

"RGList" as defined in the package limma containing data from CpG island DNA methylation array data of a neuro-ectodermal cell line that was treated with a demethylating agent. The element "weights" of the "RGList" contains the subset of invariant fragments, those without methylation-sensitive restriction sites, as a logical vector.

Details

The data is extracted from a larger experiment described in van Iterson et al. Because the data is from a high-dense tiling array a random subset of the data was chosen for convenience in making the vignette.

References

van Iterson M, Duijkers FA, Meijerink JP, Admiraal P, van Ommen GJ, Boer JM, van Noesel MM, Menezes RX (2012). A novel and fast normalization method for high-density arrays. SAGMB, 11(4).

Examples

data(methylation)

---

Functions for single-colour microarray data normalization using the P-splines

Description

Modified version of normalize.loess and normalize.AffyBatch.pspline from the affy package uses the P-spline smoother in stead of the loess algorithm

Usage

normalize.pspline(mat, epsilon = 10^-2, maxit = 1, log.it = TRUE,
    verbose = TRUE, weights = rep(1, nrow(mat)), ...)
normalize.AffyBatch.pspline(abatch,
    type=c("together","pmonly","mmonly","separate"), ...)

Arguments

mat	a matrix with columns containing the values of the chips to normalize.
abatch	an AffyBatch object.
epsilon	a tolerance value (supposed to be a small value - used as a stopping criterion).
maxit	maximum number of iterations.
log.it	logical. If TRUE it takes the log2 of mat
verbose	logical. If TRUE displays current pair of chip being worked on.
weights	For weighted normalization. The default is NULL, so there are no weights used.
type	A string specifying how the normalization should be applied. See details for more.
...	Graphical parameters can be supplied.

Details

This function is a modified version of the function normalize.loess from the affy package. In stead of the loess algorithm the function uses the P-spline algorithm. The type argument should be one of "separate","pmonly","mmonly","together" which indicates whether to normalize only one probe type(PM,MM) or both together or separately.

Value

Normalized AffyBatch

Author(s)

Maarten van Iterson and Chantal van Leeuwen

References

Laurent Gautier, Leslie Cope, Benjamin M. Bolstad and Rafael A. Irizarry (2004). affy -analysis of Affymetrix GeneChip data at the probe level. Bioinformatics, Vol. 20, no. 3, 307-315.

van Iterson M, Duijkers FA, Meijerink JP, Admiraal P, van Ommen GJ, Boer JM, van Noesel MM, Menezes RX (2012). A novel and fast normalization method for high-density arrays. SAGMB, 11(4).

Paul .H.C. Eilers and Brain D. Marx (1996). Flexible smoothing with B-splines and Penalties. Statistical Science, Vol 11, No. 2, 89-121.

See Also

normalize.loess

Examples

library(affydata)
data(Dilution)
PM <- log2(pm(Dilution[,c(1,3)]))
M <- PM[,1]-PM[,2]
A <- 0.5*(PM[,1]+PM[,2])

nPM <- log2(normalize.pspline(pm(Dilution[,c(1,3)])))
nM <- nPM[,1]-nPM[,2]
nA <- 0.5*(nPM[,1]+nPM[,2])

par(mfcol=c(2,1))
plot(M~A)
plot(nM~nA)

norm <- normalize.AffyBatch.pspline(Dilution, type="pmonly")

weights <- rep(1, nrow(exprs(Dilution)))
normw <- normalize.AffyBatch.pspline(Dilution, type="pmonly", weights=weights)

---

Panel function for adding a P-spline smoothed curves to a lattice graphics panel

Description

The function panel.pspline is similar to panel.loess but show the P-spline smoothed curve.

Usage

panel.pspline(x, y, weights = rep(1, length(y)), nintervals = 100, type, horizontal = FALSE, col.line=1, lty=1, lwd=1, ...)

Arguments

x, y	vectors giving the coordinates of the points in the scatter plot
weights	vector of weights of with same length as the data for a weighted smoothing. Default all weights are 1.
nintervals	an integer indicating the number of intervals equal to 1 + number of knots. Currently the intervals must be langer than 10.
type	see panel.loess
horizontal	see panel.loess
col.line, lty, lwd	line colour, type and width that will be used in the plots, defaults are col=1, lty=1 and lwd=1.
...	see panel.loess

Details

?panel.loess

Author(s)

Maarten van Iterson and Chantal van Leeuwen

References

Deepayan Sarkar (2009). lattice: Lattice Graphics. R package version 0.17-26. http://CRAN.R-project.org/package=lattice

van Iterson M, Duijkers FA, Meijerink JP, Admiraal P, van Ommen GJ, Boer JM, van Noesel MM, Menezes RX (2012). A novel and fast normalization method for high-density arrays. SAGMB, 11(4).

Paul .H.C. Eilers and Brain D. Marx (1996). Flexible smoothing with B-splines and Penalties. Statistical Science, Vol 11, No. 2, 89-121.

See Also

panel.loess

Examples

library(marray)
library(lattice)
data(swirl)
data <- data.frame(M=as.vector(maM(swirl)), A=as.vector(maA(swirl)), Sample=rep(paste("Array", 1:4), each=nrow(swirl)))

xyplot(M~A|Sample, data=data,
                        panel = function(x, y) {
            panel.grid(h=-1, v= 2)
            panel.xyplot(x, y)
            panel.loess(x, y, span=0.25, col="black")
                                                panel.pspline(x, y, col="red", lwd=2)})

---

Function for two-colour microarray data normalization using the P-splines

Description

Wrapper function for two colour microarray data normalization using the P-spline smoother suitable for a RGList- or MarrayRaw-objects.

Usage

pspline(object, background = c("none", "substract"), weights = NULL, nintervals = 100, subset=NULL, showArrays = 0, verbose=FALSE, line.col=2,  line.lty=1, line.lwd=2, ...)

Arguments

object	either a RGList or an MarrayRaw-object.
background	for background substraction use 'substract'. Default is no backgroud substraction.
weights	vector of weights that will be used a for a weighted normalization. The default NULL assumse equal weight 1 for all data points.
nintervals	number of bins in which the data will be divided. The default is 100 bins.
showArrays	either a integer( > 0) or a vector of integers indicating the arrays for which a MA-plot will be produced.
subset	subset of the data on which the normalization will be based. A special case of weighted normalization.
verbose	if TRUE gives additional information on the fit

.

line.col, line.lty, line.lwd	line colour, type and width that will be used in the plots, defaults are col=2, lty=1 and lwd=2.
...	additional graphical arguments for plotting.

Details

if necessary?

Value

The value that will be returned is either a MAList or MarrayNorm-object dependening on the input type.

Author(s)

Chantal van Leeuwen and Maarten van Iterson

References

van Iterson M, Duijkers FA, Meijerink JP, Admiraal P, van Ommen GJ, Boer JM, van Noesel MM, Menezes RX (2012). A novel and fast normalization method for high-density arrays. SAGMB, 11(4).

Paul .H.C. Eilers and Brain D. Marx (1996). Flexible smoothing with B-splines and Penalties. Statistical Science, Vol 11, No. 2, 89-121.

See Also

normalizeWithinArrays, maNormMain

Examples

library(marray)
data(swirl)

x <- pspline(swirl, showArrays=2, pch=20, col="grey")
x <- pspline(swirl, showArrays=2:4, line.col="green")

---

turbotrend: a fast scatterplot smoother

Description

A fast scatterplot smoother based on B-splines with second order difference penalty

Usage

turbotrend(x, y, w = rep(1, length(y)), n = 100, lambda=10^seq(-10, 10, length=1000), iter=0, method=c("original", "demmler"))

Arguments

x, y	vectors giving the coordinates of the points in the scatter plot.
w	vector of weights of with same length as the data for a weighted smoothing. Default all weights are 1.
n	an integer indicating the number of intervals equal to 1 + number of knots. Currently the intervals must be langer than 10.
lambda	Optionally a user-defined penalty parameter can be provided, if not generalized cross-validation is used to find the optimal penalty parameter.
iter	Number of robustifying iterations similar as lowess.
method	method for solving the system of linear equations either using the data in the original space or transformed to the Demmler-Reinsch basis.

Details

some details about implementation

Value

An object of type pspline is returned as a list with the following items:

x	original data vector x
y	fitted y-values with same length as vector x
w	vector of weights
n	number of bins
ytrend	binnend fitted y-values
xtrend	binned x-values
lambda	if scalar penalty parameter used else if vector of two lower and upper bound of the grid
iter	number of robustifying iterations
gcv	generalized cross-validation
edf	effective degrees of freedom (trace of the smoother matrix)
call	function call which produced this output

Author(s)

Maarten van Iterson, Chantal van Leeuwen

References

van Iterson M, Duijkers FA, Meijerink JP, Admiraal P, van Ommen GJ, Boer JM, van Noesel MM, Menezes RX (2012). A novel and fast normalization method for high-density arrays. SAGMB, 11(4).

Paul .H.C. Eilers and Brain D. Marx (1996). Flexible smoothing with B-splines and Penalties. Statistical Science, Vol 11, No. 2, 89-121.

See Also

loess,lowess, smooth, smooth.spline and smooth.Pspline

Examples

library(marray)
data(swirl)

x <- maA(swirl)[,1]
y <- maM(swirl)[,1]
xord <- x[order(x)]
yord <- y[order(x)]

plot(xord, yord, main = "data(swirl)  &  smoothing splines + lowess")
lines(turbotrend(xord, yord), col = "red", lwd=2)
lines(smooth.spline(xord, yord), col = "green", lwd=2)
lines(lowess(xord, yord), col = "purple", lwd=2)
legend("topleft", c("piecewise constant P-splines", "Cubic B-splines", "lowess"), text.col=c("red","green","purple"), bty="n")

---