Starch fine structure predicts glycemic index variation in whole-grain rice

Here, we profiled a diverse panel of whole-grain rice accessions (n = 384 covering wide range of pigmentation) for in vitro glycemic index (GI), resistant starch (RS), digestible carbohydrate (DC), and debranched starch chain-length distributions (CLD) resolved into fine degree-of-polymerization (DP) intervals. Across the panel, low-GI phenotypes were rare, and GI distributions overlapped substantially across pigmented and non-pigmented groups, indicating starch architecture as the dominant determinant of digestibility. Regression and classification models using DP-resolved predictors achieved robust GI prediction (R² = 0.70 for whole grain), and model simplification identified a reduced set of informative DP windows. Notably, DP33-36 emerged as a negative predictor of GI, showing an opposing effect relative to adjacent mid-chain intervals. To provide structural context for interval-specific effects, representative lines with contrasting DP architectures were examined by X-ray diffraction (XRD) and solid-state ¹³C NMR. Biophysical analyses supported that glycemic variation is not explained by crystalline polymorph type alone, but by localized microstructural organization within an A-type framework. For polished rice, incorporating RS content further improved the model's explanatory power (R² = 0.78). These results establish a DP-resolved structure-function framework for GI variation in rice to accelerate screening and selection of low-GI donors for breeding.