waveTiling (version 1.14.0)

wfm.fit: Fit Wfm model to trancriptome data

Description

Main function to fit a wavelet-based functional model to the tiling array expression data.

Usage

wfm.fit(object, filter.overlap=NULL, design=c("time","circadian","group","factorial","custom"), n.levels, factor.levels=NULL, chromosome, strand, minPos, maxPos, design.matrix=NULL, var.eps=c("margLik","mad"), prior=c("normal","improper"), eqsmooth=FALSE, max.it=20, wave.filt="haar", skiplevels=NULL, trace=FALSE, save.obs=c("plot","regions","all"))

Arguments

object

object of class WaveTilingFeatureSet

filter.overlap

object of class mapFilterProbe

design

character indicating the design of tiling array experiment. Currently, the following designs are implemented: time for a time-course design based on polynomial contrasts; circadian for circadian rhythm analysis; group for unordered one-factor designs; factorial for two-factor designs; custom for other designs. When using design="custom" a specific design.matrix needs to be given.

n.levels

number of levels in wavelet decomposition (integer)

factor.levels

factor levels in case of two-factor analysis. Vector of integers with length the number of factors in the experiment, and with elements the number of levels for the respective factors.

chromosome

numeric to indicate the chromosome associated with transcriptome data to fit

strand

character to indicate the strand orientation associated with transcriptome data to fit. Either "forward" or "reverse".

minPos

integer to indicate minimum genomic position

maxPos

integer to indicate maximum genomic position

design.matrix

custom design matrix to use

var.eps

character indicating how to estimate residual variance. Either "margLik" for marginal maximum likelihood based estimation or "mad" for estimation based on the MAD (more info see references).

prior

character indicating which prior distribution to put on effect functions. Either "normal" for a normally distributed prior, or "improper" for an improper prior (more info see references).

eqsmooth

logical indicating whether to force equal amount of smooth for different effect functions or not

max.it

integer giving the maximum number of iteration for estimation

wave.filt

character indicating which wavelet filter to use. Default is "haar".

skiplevels

integer indicating how many wavelet levels to put equal to 0

trace

logical indicating whether to trace estimation

save.obs

character to indicate which output to store in return object. Either "plot": all info needed to make the plots or "all": store all possible info.

Value

References

[1] Clement L, De Beuf K, Thas O, Vuylsteke M, Irizarry RA and Crainiceanu CM. (2012) Fast wavelet based functional models for transcriptome analysis with tiling arrays. Statistical Applications in Genetics and Molecular Biology 11: Iss. 1, Article 4.

[2] De Beuf K, Andriankaja, M, Thas O, Inze D, Crainiceanu CM and Clement L (2012) Model-based analysis of tiling array expression studies with flexible designs. Technical document.

Examples

  library(waveTilingData)
  data(leafdevBQ)
  data(leafdevMapAndFilterTAIR9)
  leafdevFit <- wfm.fit(leafdevBQ,filter.overlap=leafdevMapAndFilterTAIR9,design="time",n.levels=10,chromosome=1,strand="forward",minPos=22000000,maxPos=24000000,var.eps="marg",prior="improper",skiplevels=1,save.obs="plot",trace=TRUE)