Applications of the SDL Global Optimization Method in DNA Microarray Data Analysis - DNA Microarray Technology and Data Analysis in Cancer Research - page 98

Table 5.13. Genes that appear more than once in the leukemia class predictor.

Gene accession

Rank Gene IDs Frequency number Gene description

1 2642 16 U05259_rna1 MB-1 gene

2 4050 16 X03934 GB DEF = T-cell antigen receptor

gene T3-delta

3 2020 6 M55150 FAH Fumarylacetoacetate

4 1 6 AFFX-BioB-5 AFFX-BioB-5_at (endogenous

control)

5 3565 4 U66048 Clone 161455 breast expressed

mRNA from chromosome X

6 7129 4 Z78285_f GB DEF = mRNA (clone 1A7)

7 2348 3 M91432 ACADM Acyl-Coenzyme A

dehydrogenase, C-4 to C-12
straight chain

8 4991 3 Y09615 GB DEF = Mitochondrial

transcription termination factor

9 2852 2 U18004 HSU18004 Homo sapiens cDNA

10 3056 2 U32944 Cytoplasmic dynein light chain 1

(hdlc1) mRNA

11 5826 2 HG3125- Estrogen receptor (Gb:S67777)

HT3301_s

12 501 2 D50931 KIAA0141 gene

13 714 2 D87443 KIAA0254 gene

14 2327 2 M88282 T-cell surface protein tactile

precursor

15 1250 2 L08424 Achaete scute homologous protein

(ASH1) mRNA

34 test samples could be classified correctly. In contrast, the SDL
global optimization method classified all of them correctly. Moreover,
the SDL method, by selecting sets of genes based on their joint ability
to discriminate, can identify genes that are important jointly, but do not
discriminate individually. This indicates that the SDL method has
potential in identifying genes that not only discriminate between ALL
and AML, but also distinguish existing subtypes without applying any
prior knowledge.