s 12.41, which was much greater than 1.84 (the threshold), hence

st identified it as a bimodal gene as well. However, the LR test p

s one.

rozole data analysis

zole data (GDS3116) was used to examine breast cancer therapy

ness using the letrozole [Miller, et al., 2007; Miller, et al., 2010].

was to investigate if the changes of tumour expression profiling

ociated with the short-term therapy. In total, 58 patients were

sing letrozole. Microarray analysis was carried out on RNA

from these patients. Samples were paired. The samples taken

e therapy were named as the baseline samples and the samples

days after the therapy were named as the letrozole samples. Miller

alysed their data using the modified t test [Tusher, et al., 2001;

004]. They found no single gene showing consistently responding

rapy. Most genes had partially changed the expressions after the

i.e., only a subset of the patients (>45) were positively (up-

ation) or negatively (down-differentiation) responding to the

A Venn diagram for comparing the LR test, the BI test and the GM test for the

imodal genes for the letrozole drug data (GDS3116).

e 6.44 shows the Venn diagram of discovered bimodal genes

BI test and the GM test. The BI test identified about 9,000

genes. As aforementioned, its identification varies with different

e GM test identified 4,726 bimodal genes. The LR test did not