g
e limma package
ma (LInear Models for MicroArrays) package [Smyth, 2004] is
e best ones so far for employing the modified t test for a gene
n data generated using the microarray technology. In limma, a
tical approach called the shrinkage approach was used to estimate
ackage was developed by Gordon Smyth and his colleagues and
included in the Bioconductor environment. The package has been
sed to discover DEGs for the microarray gene expression data
et al., 2019; Nilson, et al., 2020]. It has also been used to analyse
gene expression data although this kind of data has a different
ibution [Mou, et al., 2020].
mma, the p values are calculated using an approach called the
Bayes (eBayes) [Smyth, 2004; McCarthy and Smyth, 2009]. By
e shrinkage approach, the standard deviations of genes are
d by the package making the extreme values move toward centre
ribution. From this, significantly differentially expressed genes
greater fold change values.
e the limma package, especially the eBayes function of the
to discover DEGs based on the microarray gene expression data,
pression matrix is required to be generated to have genes in rows
les in columns.
irst thing limma does is to generate a linear model. To do so, a
atrix is required, which defines the experimental conditions, i.e.,
cates distribute in two experimental conditions. A prostate cancer
data set (GSE3325) [Varambally, et al., 2005] was used for the
ation. The data was composed of six primary cancer samples as
x metastasis cancer samples. The code shown below was used to
a design matrix which is shown in Table 6.1.
D=data.frame(Primary=rep(1,12),
Metastasis=c(rep(0,6),rep(1,6)))