mixtNB (version 1.0)

wald.test: Wald test for performing DE analysis

Description

This function implements the Wald test for performing DE according to three statistics: difference, ratio and logratio

Usage

wald.test(out, statistic = "diff", quiet = FALSE, alpha = 0.01)

Arguments

out

The fit of mixtNB

statistic

The statistic to be used: "diff" (difference, the default), "ratio" and "logratio"

quiet

Logical to indicate if the DE genes should be printed

alpha

the significance level to detect DE genes

Value

stat

The value of the Wald test

pvalue

nominal p-values for each gene

pvalueadj

adjusted p-values according to the Benjamini and Hochberg (1995)

var

estimated variances of the genes

gname

Positions of the filtered genes

Details

This function implements the Wald test for performing DE according to three statistics: difference, ratio and logratio. It returns the statistics, the p-values and the adjusted p-values according to the Benjamini and Hochberg (1995)

References

E. Bonafede, F. Picard, S. Robin and C. Viroli (2015), Modelling overdispersion heterogeneity in differential expression analysis using mixtures, under revision.

Examples

lambda.de<-matrix(runif(100,0,250),100)
lambda.de=cbind(lambda.de,lambda.de/exp(rnorm(100,0.5,0.125)))
lambda<-rbind(lambda.de,matrix(runif(900,0,250),900,2))
a<-runif(1000,0.5,600)
cr<-rep(1:2,each=5)
y<-matrix(0,1000,10)
for (i in 1:1000) for (l in 1:10) y[i,l]<-rnbinom(1,mu=lambda[i,cr[l]],size=a[i])
fit=fit.mixtNB(y,cr,K=3)
DE.genes=wald.test(fit)