Hyperspectral phenotyping reveals the genetic basis of grain quality in rice.

Abstract

Rice ( Oryza sativa ) grain quality is an important breeding target, yet its genetic basis remains incompletely understood. In this study, we integrated hyperspectral phenotyping with genome-wide association study (GWAS) to investigate apparent amylose content (AAC) and protein content (PC) in 241 modern rice varieties. Using a visible-shortwave infrared hyperspectral system combined with optimized preprocessing and machine-learning pipelines, we achieved accurate predictions for AAC ( R 2 = 0.97) and PC ( R 2 = 0.92). Hyperspectral-based GWAS identified both known loci and previously unreported genetic associations. For AAC, qAAC (780.791nm) -1-3 was mapped to the Green Revolution gene SD1 , showing that the sd1 allele increases AAC while conferring high yields. For PC, we identified qPC (1998.98nm) -5-1 and confirmed GW5 as the causal gene, linking the high-yielding gw5 allele with high grain PC. Hyperspectral features outperformed traditional measurements, enhancing the detection of genetic signals. This study provides an efficient strategy for elucidating the genomic architecture of complex grain-quality traits.

Plant phenotyping relevance

イネ穀粒の品質形質を対象に、ハイパースペクトル計測、前処理、機械学習による形質推定を開発・評価しており、フェノタイピング手法が研究の中心である。

Code and data availability

The paper describes hyperspectral phenotyping of 241 rice varieties with GWAS, but provides no public repository, dataset URL, image deposit, or author analysis code. The data availability statement only points to the paper and its supplementary files, with no explicit statement that raw hyperspectral data, code, or模型的

No evidence-backed public reproduction asset is currently recorded.