n illustration of the distribution of the base two logarithm fold changes for the

ug data.

e statistical significance

ine whether a variable such as a gene displays statistically

nt differential expressions across two experimental conditions,

is a traditional approach [Bland, 1995]. A null hypothesis is

before running a t test,

࣢଴: ߤ஺ൌߤ஻

(6.3)

test will be used to test this hypothesis using the following

namely the t statistic [Bland, 1995], where ݑො஺ and ݑො஻ are the

d mean expressions of two experimental conditions, ߪሺݔ஺, ݔ஻ሻ is

d standard deviation of two experimental conditions,

ݐൌ^{ݑො஺െݑො஻}

ߪሺݔ஺, ݔ஻ሻ

(6.4)

alue will be derived from a t statistic and is used for the statistical

nce analysis. It will provide a statistical evidence for accepting or

the null hypothesis. Figure 6.2 shows two genes in the letrozole

[Miller, et al., 2007; Miller, et al., 2010]. The expressions of the

EK2 show a significant different distribution between two

s and hence its t test p value is 2.282e−11. Therefore, the null

is is rejected and this gene is certainly a DEG in terms of

significance. However, the expressions of the gene SLC25A30

nditions were heavily overlapped leading to a large t test p value,

almost one. Therefore the null hypothesis cannot be rejected and

is certainly not a DEG.