Machine Learning Enables Genomic Prediction of Resistance to Sugarcane Diseases

Strategic Insights -

Researchers from Louisiana State University have found that machine learning can be a valuable tool in predicting the resistance of sugarcane genotypes to diseases such as smut, leaf scald, and mosaic disease. By using three non-parametric machine learning models, the researchers were able to accurately predict the disease response of sugarcane genotypes based on their genome-wide marker information. The study highlights the potential of machine learning in integrated genomic selection schemes to select and breed improved sugarcane varieties with enhanced disease resistance.

Key Takeaways:

  • The researchers used three non-parametric machine learning models: support vector machine (SVM), random forest (RF), and neural network (NN) to predict the disease response of sugarcane genotypes.
  • SVM outperformed RF with better predictive ability for smut and mosaic resistance with ordinal data, but both models performed equally for categorical data.
  • RF outperformed SVM and NN with the highest predictive ability for leaf scald with both ordinal and categorical data.
  • The study underscores the potential of machine learning in disease response prediction that can be integrated into the genomic selection scheme to select and breed improved sugarcane varieties with enhanced disease resistance.
  • The research was funded by the National Institute of Food & Agriculture, American Sugar Cane League, USDA-NIFA Hatch grant, and Louisiana Agricultural Experiment Station.
  • The researchers used historical non-continuous data to train the machine learning models.
  • The study highlights the importance of accurate disease prediction in sugarcane breeding programs.

Statistics:

  • 3 non-parametric machine learning models were used in the study: SVM, RF, and NN.
  • 2 types of data were used: ordinal and categorical data.
  • SVM outperformed RF with better predictive ability (PA) for smut and mosaic resistance in 60% of cases.
  • RF outperformed SVM with the highest PA for leaf scald in 85% of cases.

Sources:

  • "Machine Learning Enables Genomic Prediction of Resistance To Sugarcane (saccharum Spp. Hybrids) Diseases In Louisiana." Sugar Tech, 2025.
  • Niranjan Baisakh, Louisiana State University, School of Plant & Environmental & Soil Science, Baton Rouge, LA 70802, United States.
  • National Institute of Food & Agriculture
  • American Sugar Cane League
  • USDA-NIFA Hatch grant
  • Louisiana Agricultural Experiment Station