A Novel Method for DNA Microarray Data Analysis: SDL Global Optimization Method - DNA Microarray Technology and Data Analysis in Cancer Research - page 38

microarray technology, heterogeneous cancers can be classified into
appropriate subtypes (Schena et al ., 1995). Many different kinds of
machine learning and statistical methods have recently been applied to
analyze gene expression data (Alizadeh et al ., 2000; Brown et al ., 2000;
Deutsch, 2003; Khan et al ., 2001). However, our experience tells us that
there is a need to analyze the expression of all genes detected by DNA
microarray technology at the same time, as genes always work in groups
but not individually. So far, a method that allows simultaneous analysis of
more than 30 000 genes is still lacking. Here, we introduce our novel ana-
lytical method for microarray data, using our leukemia study as an exam-
ple. This method is called SDL global optimization.
4.1. Research Subjects
Human Philadelphia chromosome-positive (Ph + ) leukemia is one of the
most commonly occurring blood cancers, and accounts for about 20% of
all leukemias. While the new therapeutic drug Gleevec is effective in
treating Ph + patients, its effectiveness in some patients is compromised
due to the development of clinical drug resistance. Improved therapy will
depend on identifying key genes that play significant roles in leukemia
development. Using our novel SDL-optimization method for gene classi-
fication and other soft computing strategies, one can perform accurate and
efficient large-scale analysis of gene expression data to identify genes
affected in leukemia cells, and the identified genes will have the potential
to serve as diagnostic markers and therapeutic targets for leukemia
patients. It will also allow monitoring disease progression during treat-
ment to reduce debilitating side effects. Moreover, strategies learnt from
our leukemia research can be applied to patients with other types of can-
cers. Developing novel techniques for improving early diagnosis using
microarray data will be at the cutting edge, and considerable potential
technology transfer opportunities may arise from the outcomes.
As stated in Chapter 2, the BCR-ABL oncogene is the cause of Ph +
leukemias. The BCR gene, on chromosome 22, breaks at either exon 1,
exon 12/13, or exon 19; and fuses to the c-ABL gene on chromosome 9 to
form, respectively, three types of the BCR-ABL chimerical gene: P190,
P210, or P230. Each of these three forms of the BCR-ABL oncogene is