Beyond differential expression: a machine learning approach to identify regulatory hubs in Moniliophthora roreri infection of cacao
摘要
Frosty pod rot (FPR), caused by Moniliophthora roreri, severely impacts global cacao (Theobroma cacao) production. While some cacao genotypes exhibit tolerance, the underlying mechanisms and the pathogen's ability to overcome them remain unclear. This study analyzed M. roreri transcript expression during infection of four cacao genotypes (two susceptible: Pound-7, CATIE-1000; two tolerant: CATIE-R4, CATIE-R7) using publicly available RNA-Seq data. We used t-SNE analysis to visualize global transcript expression, which highlighted a large cluster of M. roreri genes with minimal to no expression across all genotypes, and underscored the abundance of transcripts predicted to be secreted. Machine learning models accurately predicted M. roreri transcript expression in most genotypes but were less accurate for the tolerant CATIE-R4, suggesting a more complex pathogen response. Network analysis of 23 differentially expressed genes previously reported to be associated with tolerance identified potential functional modules, including stress response and amino acid metabolism. Furthermore, Bootstrap Forest modeling identified a methyltransferase (evm.model.sctg_0149_0001.18) and a heat shock protein (Hsp20, evm.model.sctg_0022_0002.80) as potential central regulatory hubs influencing the expression of many other M. roreri genes. These findings suggest that M. roreri employs a core transcriptional program for infection, with specific adaptations, potentially regulated by methylation and stress response pathways, to overcome tolerance. This study provides insights into the M. roreri-cacao interaction and identifies key genes and pathways that warrant further investigation as potential targets for novel FPR management strategies.