Quantum-assisted genomic prediction of blast resistance in rice (Oryza sativa L.) using SNP and haplotype representations
Rice blast caused by Magnaporthe oryzae threatens rice production, and genomic prediction may accelerate identification of resistant germplasm. We evaluated classical machine-learning and simulated quantum-kernel models for binary and three-class prediction of leaf blast (BL) and panicle blast (PB) resistance in 204 rice accessions with 2,120 repeated phenotypic observations. Models used 328,597 single-nucleotide polymorphisms (SNPs), 104,068 phased haplotypes, or 49,451 unphased haplotypes. Binary prediction outperformed three-class prediction. The highest binary macro-F1 scores were 0.87 for BL and 0.73 for PB. Phased and unphased haplotypes reduced predictor dimensionality by 68.3% and 85.0%, respectively, while retaining broadly comparable performance. Quantum support vector classification achieved competitive results in selected binary tasks but did not consistently exceed the strongest classical models. These findings establish a benchmark for SNP- and haplotype-based prediction of rice blast resistance and for evaluation of quantum machine learning in crop genomics.
Authors
- Michael Gomez Selvaraj (ORCID: https://orcid.org/0000-0003-2394-0399)
- Gloria M. Mosquera (ORCID: https://orcid.org/0000-0001-6622-7265)
- Angela Joseph Fernando
- Manuel Narváez (ORCID: https://orcid.org/0000-0002-8061-4365)
- Félicien Akohoue (ORCID: https://orcid.org/0000-0002-2160-0182)
- Jason Castano
- Robert Guerrero
Institutions
- International Center for Tropical Agriculture (KE)
- International Center for Tropical Agriculture (CO)
Publication Details
- Journal
- bioRxiv (Cold Spring Harbor Laboratory)
- Published
- 2026-11-19
- DOI
- https://doi.org/10.64898/2026.11.18.767717
- Primary Topic
- Plant Disease Resistance and Genetics
- Type
- preprint