논문 상세보기
Establishment of PLS prediction modeling of function compounds contents using multivariate analysis of spectral data from soybean (Glycine max L) seeds
- 언어ENG
- URLhttps://db.koreascholar.com/Article/Detail/297908
In this study we established the high throughput screening system of high functional soybean cultivars using PLS modeling from FT-IR spectral data of soybean(Glycine max L) seeds. Crude extract of 20% methanol from soybean seed powders (153 lines) were used for FT-IR spectroscopy. Total fatty acid, carotenoids, flavonoids and phenolic compounds contents from soybean seed powders were analyzed using UV-spectrum and GC analysis respectively. PCA analysis showed that 153 soybean lines formed a single clusters with a few outlier. PC score 1 and 2 represented 39.5, 16.4% of total variation, respectively. And than showed change patten from the middle to outside for PCA plot. We conducted PLS regression analysis between FT-IR spectral data and fatty acids data. Palmitic acid showed the highest regression coefficient (R=0.78). This result implied that the content of palmitic acid could be predicted from FT-IR spectral data from soybean seed powders with relatively high fidelity. PLS modeling of total carotenoids also showed regression coefficient of 0.69. Regression coefficient of total flavnoids and phenolic compounds were 0.44, 0.39, respectively. At present, we are trying to confirm the accuracy of PLS prediction modeling using targeted metabolite analysis (GC-MS, LC-MS) from predicted soybean lines. To increase the accuracy of PLS modeling, we also trying to standardization of spectroscopy and spectral data processing. Furthermore we are going to develop PLS modeling from GC-MS, LC-MS data. The PLS prediction modeling established in this study could be applied for high throughput screening of other leguminous plant.
