Study on domestication characteristics of Xinjiang apricot germplasm based on HPLC sugar-acid identification and machine learning

L.) is an important stone fruit in the world that originated in China. The flavor of apricot fruit is mainly determined by its sugar and acid components. Studies have shown that apricot landraces and traditional cultivars can be distinguished by determining the sugar and acid components of the fruit using high-performance liquid chromatography (HPLC) and machine learning approaches. Total soluble sugar, malic acid, and citric acid exhibited no significant differences between the two groups. However, correlation analysis indicated that sugar coordination was enhanced in traditional cultivars compared with landraces, while the organic acid composition was more diverse. Four machine learning models classified them with performance metrics above 0.85. UpSetR analysis identified oxalic acid, tartaric acid, and fructose as key indicators, further improving model performance vs. full features. This study provides a novel strategy based on machine learning to distinguish apricot domestication traits across periods.