188    ◾    Bioinformatics

We can plot the BCV against gene abundance (in log2 counts per million) using

“plotBCV(y)” function. As described above, the BCV is the square root of the dispersion

parameter in the negative binomial model.

jpeg(‘myBCVPlot.jpg’)

plotBCV(yNorm, pch=16, cex=1.2)

dev.off()

The above script plots the biological coefficient of variations (based on the three types of

dispersions) against the average abundance of each gene. As shown in Figure 5.14, the plot

shows the square root estimates of the common, trended, and tagwise negative binomial

dispersions.

For RNA-Seq count data, the negative binomial dispersions tend to be higher for genes

with very low counts and the dispersion trend tends to decrease smoothly with the abun-

dance (CPM) increase and becomes asymptotic to a constant value for genes with a larger

abundance.

FIGURE 5.13  Printing dispersions.

FIGURE 5.14  Biological CV plot.