Submitted to: Theoretical and Applied Genetics
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 3/24/2026
Publication Date: 4/8/2026
Citation: Ahn, E.J., Oh, S., Prom, L.K., Lim, S., Meinhardt, L.W., Magill, C. 2026. A haplotype-layered GWAS identifies a multi-trait grain mold resistance hub on sorghum chromosome 5. Theoretical and Applied Genetics. https://doi.org/10.1007/s00122-026-05222-9.
DOI: https://doi.org/10.1007/s00122-026-05222-9

Interpretive Summary: Finding the genes that protect sorghum from grain mold is like trying to read a scrambled book: looking at single letters (standard DNA markers) is confusing and often leads to errors. To solve this, we developed a new method called "Haplo-kmer" analysis that reads the genome in full "words" (haplotype blocks) rather than isolated letters. This approach allowed us to cut through the noise and identify a robust "Genetic Command Center" on Chromosome 5 that controls resistance against multiple mold symptoms simultaneously. Unlike previous studies that found weak, scattered signals, our method pinpointed a specific, stable region containing novel defense candidates validated by machine learning models, which work together as a hub. This research provides plant breeders with a reliable "search term" to identify and breed mold-resistant sorghum varieties with high accuracy, ultimately securing better harvests for farmers and safer grain for consumers.

Technical Abstract: The genetic dissection of quantitative resistance to sorghum grain mold is frequently confounded by low marker density and inconsistent SNP-trait associations. To resolve this "missing heritability" problem, we developed a unified "Haplo-kmer" GWAS framework that integrates standard SNPs with local haplotype-window markers derived from GBS data. By applying a stringent statistical threshold for multi-trait convergence across the Sorghum Association Panel (SAP), we identified a pleiotropic resistance hub on Chromosome 5 (60.25–60.50 Mb) that was fragmented or undetectable in single-SNP analysis. Haplotype-based mapping demonstrated superior signal stability in rank-preserving permutations and revealed a specific resistant haplotype block linked to candidate genes Sobic.005G141700 and Sobic.005G142000, the former being independently prioritized by non-linear machine learning models. This study validates that local haplotype structure captures epistatic or structural variants often missed by linear models, establishing the Chromosome 5 hub as a critical, stable target for marker-assisted selection in grain mold resistance breeding.