Using Network Extracted Ontologies to Identify Novel Genes with Roles in Appressorium Development in the Rice Blast Fungus Magnaporthe oryzae

Abstract

Magnaporthe oryzae is the causal agent of rice blast disease, the most important infection of rice worldwide. Half the world's population depends on rice for its primary caloric intake and, as such, rice blast poses a serious threat to food security. The stages of M. oryzae infection are well defined, with the formation of an appressorium, a cell type that allows penetration of the plant cuticle, particularly well studied. However, many of the key pathways and genes involved in this disease stage are yet to be identified. In this study, I have used network-extracted ontologies (NeXOs), hierarchical structures inferred from RNA-Seq data, to identify pathways involved in appressorium development, which in turn highlights novel genes with potential roles in this process. This study illustrates the use of NeXOs for pathway identification from large-scale genomics data and also identifies novel genes with potential roles in disease. The methods presented here will be useful to study disease processes in other pathogenic species and these data represent predictions of novel targets for intervention in M. oryzae.

Keywords: Magnaporthe oryzae; RNA-Seq; fungal pathogen; gene expression; network; plant pathogen; rice blast.

Figure 1

Distribution of correlations for all pairs of M. oryzae genes. Pearson’s Correlation Coefficient (R${}^2$) are reported as absolute values.

Figure 2

The structure of a M. oryzae network-extracted ontology. Nodes represent entities that contain groups of genes and edges represent relationships between entities. Entities are sized by the number of genes they contain and edges are shaded to show a measure of distance between entities. Entities coloured in blue and red depict clusters of genes that are enriched for up-regulated genes at 4–8 h and 14–16 h timepoints during appressorium development respectively. Entities that contain five or more genes are included in the visualisation. The root node that contains the whole system is labelled.

Figure 3

Correlation of entity size and robustness score. Entity robustness is calculated based on network support from co-expression data and from bootstrap analysis. There is a strong positive correlation between entity size and robustness (R = 0.84, p≪$0.0001$).

Figure 4

Examples of entities enriched for up-regulated M. oryzae genes during appressorium development. CLX0008701 and CLX0008712 are enriched for genes during both 4–8 h and 14–16 h of appressorium development. Entities CLX0008712 and CLX0008695 are enriched for genes up-regulated during 14–16 h of appressorium development. Nodes represent genes and edges between nodes represent correlated expression of genes across 46 RNA-Seq samples. Edges are coloured by Pearson’s Correlation Coefficient (R${}^2$) and only shown if R${}^2$> 0.9. Genes up-regulated during 4–8 h and 14–16 h of appressorium development are coloured blue and red respectively.