Computational identification of functional network modules associated with the pathogenicity of Fusarium verticillioides

Biological functions are executed through elaborate collaboration of various biomolecules, and there has been increasing interest in the computational identification of functional modules from large-scale experimental data. In this study, we performed a comparative analysis of two distinct Fusarium verticillioides RNA-Seq datasets, where one set was obtained from wild-type F. verticillioides and the other set from a loss-of-virulence mutant. For a systematic analysis of the infection transcriptome, we first predicted the co-expression network of the fungus. Subsequently, we identified functional subnetwork modules in the co-expression network consisting of interaction genes that display strongly coordinated behavior in the respective datasets. A probabilistic pathway activity inference method was adopted to identify modules likely to be involved in the pathogenicity of F. verticillioides, and a computationally efficient branch-out technique was used to search for potential subnetwork modules. Our results revealed four potential subnetwork modules, where the modules contained several enriched GO terms as well as potential pathogenic genes that are orthologous to known pathogenic genes in other fungi.