DNA microarray data analysis for cancer classification based on stepwise discriminant analysis and Bayesian decision theory

摘要

DNA microarray are providing an unprecedented amount of information about the genetic changes. From the analysis of the data, we can get insights into various diseases such as cancer whose information is difficult to be obtained. Golub et al. showedthe possibility of cancer classification based on gene expression profiles. They clearly showed that the detection of wholesale changes in gene expression patterns is possible. However, researchers could not identify the genes whose activity is turned up or down. They also could not find which changes of genes are important for cancer development and progression-the causes and not just the effects. Therefore, the useful methods to analyze DNA rnicroarray data have not been proposed. In this paper, we find the causal relationship between several tumors and the gene-expression data by sequentially using the stepwise discriminant analysis method (SDA) and Bayesian decision theory (BDT). Eighty-five samples containing four tumor classes are used in this study. The classes are neuroblastoma (NB), rhabdomyosarcoma (RMS), non-Hodgkin lymphoma (BL) and the Ewing family of tumor (EWS). These data are open on world wide web. SDA is used to select critical genes for accurate classification of 4 tumors from original 2308 genes. With the selected genes, Bayesian classifier is made, which minimizes the misclassification rate. As a result, the classification performance increases to 100 percent, and 9 new genes that have relation with the development of the tumorsis found additionally.