he distribution of the errors between the raw delta expressions and the predicted

ssions for genes in the letrozole drug data.

use expression levels can vary with genes, fitting a Gamma

unction to each individual gene may have a significant variation

lta expression significances across genes. Such a heterogeneous

may lead to inconsistent interpretation of the discovered bimodal

a data set. In order to provide more robust estimation of the delta

n significance, a Gamma density function was fitted to whole

delta expressions for the letrozole drug breast cancer treatment

ure 6.36 shows the result. It can be seen that such an approach

ntly reduced the fitting error, i.e. from 0.3664 to 0.000252.

e, fitting a single Gamma density function to whole genome delta

n may provide a more robust way to evaluate the delta expression

nce gene-wise, and hence very useful for gene expression

ty pattern discovery.

he Gamma density function fitted to the delta expressions for the letrozole drug