Machine learning-driven GWAS uncovers novel candidate genes for resistance to frosty pod rot and witches' broom disease in cacao

Authors: Ahn, Ezekiel; Park, Sunchung; Baek, Insuck; LEE, DONGHO - Orise Fellow; BHATT, JISHNU - Orise Fellow; LIM, SEUNGHYUN - Orise Fellow; Jang, Jae Hee; Zhang, Dapeng; Kim, Moon; Meinhardt, Lyndel

Submitted to: The Plant Genome
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 6/2/2025
Publication Date: 7/11/2025
Citation: Ahn, E.J., Park, S., Baek, I., Lee, D., Bhatt, J., Lim, S., Jang, J., Zhang, D., Kim, M.S., Meinhardt, L.W. 2025. Machine learning-driven GWAS uncovers novel candidate genes for resistance to frosty pod rot and witches' broom disease in cacao. The Plant Genome. 18(3). Article e70069. https://doi.org/10.1002/tpg2.70069.
DOI: https://doi.org/10.1002/tpg2.70069

Interpretive Summary: Cacao, the source of chocolate, is a vital crop for the global confectionary industry and the livelihoods of millions of farmers. However, cacao production is severely threatened by diseases like frosty pod rot (FPRD) and witches' broom disease (WBD), which can devastate yields. This study employed advanced machine learning techniques to analyze the genetic makeup of over 100 diverse cacao trees and identify genes associated with resistance to these diseases, as well as healthy pod production. We discovered a suite of genes potentially involved in fortifying the plant's cell walls, activating stress responses, and deploying defense mechanisms. These findings highlight the complex genetic architecture of disease resistance in cacao, revealing that many genes, each with a small effect, likely work together to protect the plant. This research provides valuable insights for breeding programs aiming to develop cacao varieties that are more resilient to FPRD and WBD. By understanding which genes are involved in disease resistance and how they interact, breeders can more effectively select and combine these genes to create superior varieties. Ultimately, this work contributes to the long-term sustainability of cacao production, ensuring a stable supply of chocolate and supporting the livelihoods of cacao farmers worldwide.

Technical Abstract: This study employed machine learning-driven genome-wide association studies (GWAS) to dissect the genetic architecture of resistance to frosty pod rot (FPRD) and witches' broom disease (WBD), as well as healthy pod rate, in a diverse collection of 102 cacao (Theobroma cacao) accessions. Phenotypic data for these traits were analyzed in conjunction with SNP markers mapped to both the Criollo and Matina reference genomes. Bootstrap Forest and Boosted Tree models were used to identify significant SNP-trait associations. Our analyses revealed numerous candidate genes associated with disease resistance and productivity, highlighting the polygenic nature of these traits. Identified genes are implicated in various biological processes, including cell wall biosynthesis and modification, biotic and abiotic stress response signaling, and defense-related mechanisms such as RNA silencing and ion transport. Notably, associations varied depending on the reference genome used, emphasizing the complexity of the cacao genome and the importance of considering genomic structural variation in GWAS. This research provides a genomic roadmap for developing cacao varieties with enhanced resistance to FPRD and WBD, contributing to the sustainability of cacao production. The application of machine learning methods in this study demonstrates their power in uncovering complex genetic interactions in polygenic traits and provides a framework for future genomic studies in cacao and other crops.