随机矩阵理论在肺癌基因网络识别中的应用

利用随机矩阵理论（RMT）方法除去肺癌基因表达数据中的噪声，并将去噪后的数据分别用模块方法和等级聚类方法进行处理.比较两种方法处理后的结果，发现RMT-等级聚类方法不仅能给出模块，还能给出模块间的关联强度.研究表明，RMT-等级聚类方法是一种有效的识别基因网络的新方法. 关键词： 随机矩阵理论 等级聚类 基因网络 肺癌

Application of random matrix theory to identification of lung cancer gene networks

We used random matrix theory （RMT） to remove the noises in lung cancer gene expression data and used the modules approach and the hierarchical clustering approach to construct the gene networks. Comparing the results given by the two methods, we found that RMT-hierarchical clustering method gives true modules as well as the correlations between the modules. The results indicate that RMT-hierarchical clustering method is an effective new method for identifying gene networks.