Abstract
Gene coexpression analysis that depicts gene-to-gene correlations became prevalent in plant transcriptome research. Methods of network analysis have been developed to detect connected groups of genes from a large network. However, within the group, it was difficult to exclude the possibility that the group was also highly connected with genes on the outside of the group. To extract genes with dense within-group connectivity and sparse outside connectivity, we developed a novel algorithm by introducing quantitative index "network specificity". To identify genes specifically coexpressed with a query gene, the first procedure was to extract a group of genes that have high network specificity to a query gene. The second procedure was to calculate network specificity of the group to estimate within-group coexpression quantitatively. By applying this algorithm to genome-wide analysis of Arabidopsis gene coexpression, we identified a number of coexpression groups.
