ay be different.

delta expression of a gene was generated using the following

. Suppose a vector of K (58 for the letrozole drug data) case

ns was expressed as below, where n was the n^th gene in data,

ܠ௡ൌ൫ݔ௡,ଵ, ݔ௡,ଶ, ⋯, ݔ௡,௄൯

(6.38)

xpressions in this vector were sorted in an ascending order and

d as below,

ܠ௡ൌ൫ݔ௡,ଵ

ᇱ, ݔ௡,ଶ

ᇱ, ⋯, ݔ௡,௄

ᇱ

൯

(6.39)

ta expression vector was generated by calculating the difference

all consecutive pairs of sorted expressions,

ൌ൫ݔ௡,ଶ

ᇱ

െݔ௡,ଵ

ᇱ, ݔ௡,ଷ

ᇱ

െݔ௡,ଶ

ᇱ, ⋯, ݔ௡,௄

ᇱ

ᇱ

െݔ௡,ሺ௄ିଵሻ൯

(6.40)

e 6.32 has shown that the Gamma density modelling process

ery well for fitting the Gamma density to the delta expressions of

es selected from the letrozole drug data. Based on an estimated

density function, the significance of each delta expression can be

d.

e 6.33 shows the relationship between the Gamma p values and

expressions for three genes from the letrozole drug data shown

6.32. In these plots, it can be seen that some genes contained

ntly great delta expressions, such as the genes ZMIZ2 in Figure

nd ATP10B in Figure 6.33(b). However, some genes did not

a significantly great delta expression, such as the gene

P28 in Figure 6.33(c). A gene containing no such a significantly

ta expression implies that the expressions of such a gene very

m a unimodal distribution. Otherwise, the expressions of such a

y form a multi-modal distribution including a bimodal distribution

s a gene with outliers.