n pattern analysis and has been popular in the biological/medical

ysis research. It was rooted from the so-called Bland-Altman plot

or a MA plot), which is also called the Tukey mean difference

he descriptive statistics used in various applications [Bland and

1986; Bland and Altman, 1999]. Such a plot can be used to

the relationship between two meta variables, which are derived

umber of raw variables. They are the variable difference and the

mean.

asic assumption of gene expression data is that most genes will

significantly differential responses to a stress across conditions.

r, it is also assumed that many genes may have small mean

n values. In a MA plot, the x-coordinate uses the variable mean,

mean expression variable. The y-coordinate uses the variable

e, i.e., the differential expression variable. To generate a MA plot

a, a MA data structure needs to be defined in limma. The code

below, where sam.model is an eBayes model generated by

, $Amean is the mean expression variable and the differential

n variable is coefficients[,2], which is the fold change.

("MAList")

am.model$Amean

am.model$coefficients[,2]

wards, a vector of the classification of genes is defined to indicate

s of each gene, i.e., being a down-regulated DEG, or an up-

DEG or a non-DEG.

=rep('Non-DEG',nrow(X))

am.model$coefficients[,2]

.model$p.value[,2]

[which(fold< -2 & pv<0.01)]='Down'

[which(fold> 2 & pv<0.01)]='Up'

y, a limma function plotMA is called to generate a MA plot for

(MA,status=status,values=values,hl.col=col)

range(MA$A),c(-2,-2),lty=3,lwd=2)