Beyond differential expression: a machine learning approach to identify regulatory hubs in Moniliophthora roreri infection of cacao

Authors: LIM, SEUNGHYUN - Orise Fellow; Park, Sunchung; BHATT, JISHNU - Orise Fellow; Jang, Jae Hee; Zhang, Dapeng; Meinhardt, Lyndel; Ahn, Ezekiel

Submitted to: European journal of plant pathology
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 10/16/2025
Publication Date: 11/3/2025
Citation: Lim, S., Park, S., Bhatt, J., Jang, J., Zhang, D., Meinhardt, L.W., Ahn, E.J. 2025. Beyond differential expression: a machine learning approach to identify regulatory hubs in Moniliophthora roreri infection of cacao . European journal of plant pathology. https://doi.org/10.1007/s10658-025-03154-6.
DOI: https://doi.org/10.1007/s10658-025-03154-6

Interpretive Summary: Cacao, the key ingredient in chocolate, is a globally important crop supporting the livelihoods of millions of farmers. However, its production is significantly threatened by frosty pod rot (FPR), a devastating disease caused by the fungus Moniliophthora roreri. While some cacao varieties exhibit tolerance to FPR, the fungus can adapt and overcome these defenses. This study used advanced techniques, including network analysis and machine learning, to analyze the genetic responses of M. roreri during the infection of both susceptible and tolerant cacao varieties. We identified a core set of fungal genes essential for infection, regardless of the host's resistance level. Our analysis revealed two potential "master switches", that appear to control many other genes involved in the infection process. These findings offer valuable insights into how M. roreri infects cacao and adapts to host defenses. This knowledge can guide the development of new disease management strategies, such as creating cacao varieties with enhanced resistance that targets these key fungal genes. Ultimately, this research contributes to safeguarding the future of chocolate production and improving the livelihoods of cacao farmers worldwide. This research is particularly relevant to plant pathologists, fungal biologists, cacao breeders, and the broader agricultural community working towards sustainable cacao production.

Technical Abstract: This study investigated the transcriptional dynamics of Moniliophthora roreri during the infection of four Theobroma cacao genotypes, two susceptible (Pound-7, CATIE-1000) and two tolerant (CATIE-R4, CATIE-R7), using publicly available RNA-Seq data. t-SNE analysis revealed a core M. roreri transcriptome deployed across all genotypes, enriched for predicted secreted proteins. Boosted Neural Network models accurately predicted M. roreri gene expression in susceptible genotypes and the tolerant CATIE-R7 (R-square > 0.90), but were significantly less accurate for CATIE-R4 (R-square = 0.69), suggesting a more complex or distinct pathogen response to this genotype. Network analysis of 23 previously reported differentially expressed genes associated with tolerance, combined with Bootstrap Forest modeling, identified a putative methyltransferase (evm.model.sctg_0149_0001.18) and an Hsp20 heat shock protein (evm.model.sctg_0022_0002.80) as potential central regulatory hubs influencing the expression of other M. roreri genes. These findings suggest that M. roreri utilizes a core transcriptional program for infection, with specific adaptations, potentially mediated by methylation and stress response pathways, to overcome tolerance. The identification of these potential regulatory hubs provides novel targets for further investigation into their precise roles in M. roreri virulence and adaptation. This research advances our understanding of the molecular mechanisms underlying the M. roreri-cacao interaction and provides a framework for developing innovative strategies to manage frosty pod rot disease. This work will be of particular interest to researchers in fungal biology, molecular plant pathology, and those involved in cacao improvement programs.