With the help of an extension of k-means clustering method we construct cluster trees of gene expressions. Most gene is located on a specific position in an edge of the tree between two cluster centers. This information should be used for further study of interrelation and expression studies of genes. The tree shows the structure of the clusters and a possible evolution connections among different clusters of genes and it helps to identify networks of genes. The method applied for several data set and the applications will be presented in the talk.