Huiyu Gao1, Guodong Wang1, Jianhua Men1, Zhu Wang1. 1. National Institute for Nutrition and Health, Chinese Center for Disease Control and Prevention, Beijing 100050, China.
Abstract
OBJECTIVE: To explore the potential of near-infrared reflectance( NIR)spectroscopy to determine macronutrient contents in beans. METHODS: NIR spectra and analytical measurements of protein, moisture and ash were collected from 70 kinds of beans. Reference methods were used to analyze all the ground beans samples. NIR spectra on intact and ground beans samples were registered. Partial least-squares( PLS)regression models were developed with principal components analysis( PCA) to assign 49 bean accessions to a calibration data set and 21 accessions to an external validation set. RESULTS: For intact beans, the relative predictive determinant( RPD) values for protein and ash( 3. 67 and 3. 97, respectively) were good for screening. RPD value for moisture was only 1. 39, which was not recommended. For ground beans, the RPD values for protein, moisture and ash( 6. 63, 5. 25 and 3. 57, respectively) were good enough for screening. The protein, moisture and ash levels for intact and ground beans were all significantly correlated( P < 0. 001) between the NIR and reference method and there was no statistically significant difference in the mean with these three traits. CONCLUSION: This research demonstrates that NIR is a promising technique for simultaneous sorting ofmultiple traits in beans with no or easy sample preparation.
OBJECTIVE: To explore the potential of near-infrared reflectance( NIR)spectroscopy to determine macronutrient contents in beans. METHODS: NIR spectra and analytical measurements of protein, moisture and ash were collected from 70 kinds of beans. Reference methods were used to analyze all the ground beans samples. NIR spectra on intact and ground beans samples were registered. Partial least-squares( PLS)regression models were developed with principal components analysis( PCA) to assign 49 bean accessions to a calibration data set and 21 accessions to an external validation set. RESULTS: For intact beans, the relative predictive determinant( RPD) values for protein and ash( 3. 67 and 3. 97, respectively) were good for screening. RPD value for moisture was only 1. 39, which was not recommended. For ground beans, the RPD values for protein, moisture and ash( 6. 63, 5. 25 and 3. 57, respectively) were good enough for screening. The protein, moisture and ash levels for intact and ground beans were all significantly correlated( P < 0. 001) between the NIR and reference method and there was no statistically significant difference in the mean with these three traits. CONCLUSION: This research demonstrates that NIR is a promising technique for simultaneous sorting ofmultiple traits in beans with no or easy sample preparation.
Entities:
Keywords:
beans; macronutrients; near-infrared reflectance; protein