Figure 1

Overview of the proposed network-based comparative analysis pipeline for predicting potential maize subnetwork modules associated with maize defense response. 1st step: the RNA-seq data were preprocessed by aligning them to the reference genome and filtering out lowly expressed genes for quality control to obtain the gene expression matrix. 2nd step: in order to predict important candidate genes potentially involved in maize defense modules, a cointegration-correlation-expression approach was applied to identify maize genes, whose expression patterns correspond to those of selected F. verticillioides pathogenicity genes. 3rd step: co-expression networks surrounding the candidate maize genes were predicted and a log-likelihood ratio (LLR) matrix was computed for subsequent analysis. Through a seed-and-extend approach with an efficient branch-out technique, we searched for potential maize subnetwork modules starting from the top 20% differentially expressed seed genes. Finally, potential maize subnetwork modules involved in defense response were predicted by evaluating the strength of the association between the module activity level and the pathogenicity of the fungi.