Identification of essential or lethal genes would be one of the ultimate goals in drug designs. Here we introduce an in silico method to select the Cluster with a high population of lethal genes, called lethal cluster, through microarray assay. We construct a gene transcription network based on the microarray expression level. Links are added one by one in the descending order of the Pearson correlation coefficients between two genes. As the link density p increases, two meaningful link densities p(m), and p(s) are observed. At p(m), which is smaller than the percolation threshold, the number of disconnected clusters is maximum, and the lethal genes are highly concentrated in a certain cluster that needs to be identified. Thus the deletion of all genes in that cluster could efficiently lead to a lethal inviable mutant. This lethal cluster can be identified by an in silico method. As p increases further beyond the percolation threshold, the power law behavior in the degree distribution of a giant cluster appears at p(s). We measure the degree of each gene at p(s). With the information pertaining to the degrees of each gene at p(s). We return to the point p(m) and calculate the mean degree of genes of each cluster. We find that the lethal cluster has the largest mean degree. (c) 2005 Elsevier B.V. All rights reserved.