Gene Expression Data Cluster Analysis

The explosive growth of the gene expression data needs an automatic and effective data analysis tool urgently. Presently, clustering has become the powerful and widely used method in gene expression data analysis to obtain biological information. However, there are problems in analyzing gene expression data of over-dependence on the distribution of dataset and impossibly achieving a global optimal clustering effect. This paper introduces the spectral clustering method. The advantage of this method is that it can be used in any shape of sample space and converge in the global optimal. In experiment, We use yeast cell cycle and Lyer´s serum data set as the test data set and select adjust-FOM as the evaluation criteria. The result shows the spectral clustering method in the clustering effect is better than traditional clustering methods.