由資料採礦技術探討微陣列資料篩選乳癌相關基因

根據行政院衛生署2007年的統計資料顯示，在2007年國人十大死因排行，惡性腫瘤（癌症）連續26年蟬聯首位，在各種癌症之中值得注意的是女性乳癌，其發生率有逐年高升的趨勢，在2005年女性乳癌發生率高達58.83（每十萬人口），位居於第一位，對於婦女同胞的健康構成極大的威脅。目前在生物或醫學研究上，常常使用一些對乳癌有影響的已知基因或激素來進行辨別是否可能罹患乳癌，例如像p53、BRCA1、BRCA2、ER、HER-2…等等，但是，在這成千上萬種的基因裡，應該還是存在著一些可以用來進行辨別有沒有可能罹患乳癌的基因。本研究採用Affymetrix GeneChip Human Genome U133 Plus 2.0 Array這組晶片所做出來的microarray資料為材料來進行研究，所使用的資料集為NCBI裡面的GSE2109、GSE3744、GSE7307三個資料集。本研究使用了集群分析來篩選出在乳癌病患及正常個體之間表現有差異的基因，並建立判別模型來判別一個新個體資料是否可能為乳癌病患，最後再結合GO基因功能分類資料庫，將篩選出的基因進行基因功能分類，提供生物及醫學研究上之參考。

關鍵字

微陣列；乳癌；資料採礦；集群分析；判別分析

並列摘要

According to the report of leading causes of death from the Department of Health of the Executive Yuan incidence in 2007, the malignant neoplasm are the most of number of death from 1982 to 2007. Breast cancer is worth watching out in the cancer, the incidence rate of breast cancer have been greatly increasing over the past decades, it approached 58.83 per 100,000 per year, it became the threaten of female health. Researchers in medicine or biology used some known genes or hormone that affected breast cancer to identify who had breast cancer or not, like p53, BRCA1, BRCA2, ER, and HER-2, but it might have lots of genes that could make a distinction between cancer patients and normal people.We used the Affymetrix GeneChip Human Genome U133 Plus 2.0 Array, our data sets downloaded from NCBI, they were GSE2109, GSE3744, and GSE7307. We collected the genes that had significant difference in expression with cluster analysis and created a discriminant function with discriminant analysis. Finally, we found out the gene ontology terms that the genes we collected, provided some useful index for researchers in medicine or biology in the future.