MLM.beadarray: Multi-level Mixed model

In beadarrayFilter: Bead filtering for Illumina bead arrays

Description

Function for differential expression analysis of bead array data using the Multi-level Mixed model of Kim and Lin (2011).

Usage

MLM.beadarray(sig, stderr, nbeads, groups, var.equal = FALSE, max.iteration = 20,
epsilon = 1e-06, method = "REML")

Arguments

sig	The summarized and normalized average intensities
stderr	The standard errors of the means
nbeads	The number of beads used for summarization
groups	This refers to the groups to be compared. For filtering, group refers to the number of cloumns of sig. However, for the differential expression analysis, the user should define the group variable as appropriate for his/her data.
var.equal	i.e assuming equal variance for the variance of the array random effects
max.iteration	The maximum number of iteration to perform
epsilon	control limit for convergence
method	Allows one to choose between restricted maximum likelihood (REML) or maximum likelihood (ML) estimations

Author(s)

Ryung S. Kim and Juan Lin

References

Kim, R.S. and Lin, J. (2011). Multi-level mixed effects models for bead arrays. Bioinformatics, 27(5):633-640.

Examples

require(beadarrayExampleData)
data(exampleSummaryData)
exampleSummaryDatalog2 <- channel(exampleSummaryData, "G")[1:40,]

exampleSummaryDataNorm <- normaliseIllumina(exampleSummaryDatalog2, 
method = "quantile", transform = "none")[1:40,]

    eSet <- na.omit(exprs(exampleSummaryDataNorm))[1:40,]
    
    seSet <- na.omit(se.exprs(exampleSummaryDataNorm))[1:40,]
    
    nSet <- na.omit(attributes(exampleSummaryDataNorm)$assayData$nObservations)[1:40,]
    
     stderrs<-seSet/sqrt(nSet)

##define group variable as appropriate for your data
group1 <- c(1:6)
group2 <- c(7:12)
 fit1 <- MLM.beadarray(eSet, stderrs, nSet, list(group1,group2), var.equal = TRUE,
 max.iteration = 20, method = "ML")
   
df<-length(group1)+length(group2)-2
fit1$pvalue<-2*(1-pt(abs(fit1$t.statistics),df))
fit1$PvalADjust<-p.adjust(fit1$pvalue, method ="fdr", n = length(fit1$pvalue))
length(which(fit1$PvalADjust<0.05)) 

beadarrayFilter documentation built on May 2, 2019, 6:05 a.m.