Skip to Main Content
The aim of this study is to select significant genes associated with spermatogenesis. We use singular value decomposition to extract the principal component of the DNA microarray dataset and the expression profile of the first eigengene shows the expression tendency of most genes' expression. Basing on this observation, we rank genes using correlation of each gene expression and the first eigengene's profile. Five kinds of correlation methods are considered and experimental results on a real spermatogenesis microarray dataset demonstrate that the cosine correlation method can find 75 informative genes among top 100 probes/genes. The selected significant genes, especially the top 10 probes/genes, are related to spermatogenesis through literature analysis.