【科研动态】作物所在甘薯块根高通量NIRS表型分析方面取得新进展
近日，作物所甘薯团队在国际学术期刊Food Chemistry-X（中科院大类一区，IF=6.1）发表题为“High-throughput phenotyping of nutritional quality components in sweet potato roots by near-infrared spectroscopy and chemometrics methods”的文章。作物所唐朝臣博士为论文第一作者，王章英研究员为通讯作者。
甘薯（Ipomoea batatasL.）是一种高产且营养丰富的块根作物。提高块根品质是促进甘薯产业高质量发展的重要举措。然而，缺乏高效的甘薯品质分析方法严重制约了甘薯种质的品质鉴评和优质新品种的选育。为了解决这一问题，本研究选用了125份代表性的甘薯块根样品，采用了双重优化的建模策略（即样本子集划分和光谱变量选择），旨在构建一种基于NIRS的高通量分析块根品质（总淀粉、直/支链淀粉、支直比、可溶性糖、粗蛋白、总黄酮和总酚）的方法，为快速筛选优质甘薯种质提供可行的解决方案。在本研究中，共建立了8个最优的NIRS定量预测模型，校正集决定系数（R2C）为0.95–0.99，交叉验证决定系数（R2CV）为0.93–0.98，验证集决定系数（R2V）为0.89–0.96，验证集相对分析误差（RPD）为6.33–11.35。总之，本研究开发的NIRS模型为高通量分析甘薯块根品质提供了一种实用可行的方法，为实现高效、精准的甘薯品质育种奠定了基础。
本研究得到国家重点研发计划子课题、国家甘薯产业技术体系、广东省现代农业产业技术体系、广东省农业科学院作物研究所所长基金、广东省农业科学院科技人才引进/培养专项等项目的资助。
原文链接：
https://www.sciencedirect.com/science/article/pii/S2590157523003590
Fig. 1. Variations in quality components among representative sweet potato samples. (a) Higher content group, including total starch (TS), amylose (AL), amylopectin (AP), and soluble sugar (SS). (b) Lower content group, including ratio of amylopectin to amylose (RAA), crude protein (CP), total flavonoid content (TFC), and total phenolic content (TPC).
Fig. 2. Variations in NIRS absorbance spectra among representative sweet potato samples. (a) Original near-infrared spectra of hot-air-dried samples. (b) PCA scores of near-infrared spectra for hot-air-dried samples. (c) Original near-infrared spectra of freeze-dried samples. (d) PCA scores of near-infrared spectra for freeze-dried samples.
Fig. 3. The selection of variables for quality components of sweet potato samples in the calibration set using competitive adaptive reweighted sampling (CARS), random frog (RF), and Monte Carlo-uninformative variable elimination (MC-UVE). (a) TS, total starch; (b) AL, amylose; (c) AP, amylopectin; (d) RAA, ratio of amylopectin to amylose; (e) SS, soluble sugar; (f) CP, crude protein; (g) TFC, total flavonoid content; (h)TPC, total phenolic content.
Fig. 4. Predictive performance of partial least squares models for the quality components. FS, full spectra; CARS, competitive adaptive reweighted sampling; RF, random frog; R2C, coefficient determination of calibration; R2CV, coefficient determination of cross-validations; R2V, coefficient determination of validations; RMSEC, root mean standard error of calibration; RMSECV, root mean standard error of cross-validation; RMSEP, root mean square error of prediction; RPD, the ratio of prediction to deviation; RER, the range error ratio. (a) TS, total starch; (b) AL, amylose; (c) AP, amylopectin; (d) RAA, ratio of amylopectin to amylose; (e) SS, soluble sugar; (f) CP, crude protein; (g) TFC, total flavonoid content; (h)TPC, total phenolic content.