Rapid discrimination of glycogen particles originated from different eukaryotic organisms
Jia-Wei Tang,Rui Qiao,Xue-Song Xiong,Bing-Xin Tang,You-Wei He,Ying-Ying Yang,Pei Ju,Peng-Bo Wen,Xiao Zhang,Liang Wang
DOI: https://doi.org/10.1016/j.ijbiomac.2022.09.233
IF: 8.2
2022-12-01
International Journal of Biological Macromolecules
Abstract:There are many commercially available glycogen particles in the market due to their bioactive functions as food additive, drug carrier and natural moisturizer, etc. It would be beneficial to rapidly determine the origins of commercially-available glycogen particles, which could facilitate the establishment of quality control methodology for glycogen-containing products. With its non-destructive, label-free and low-cost features, surface enhanced Raman spectroscopy (SERS) is an attractive technique with high potential to discriminate chemical compounds in a rapid mode. In this study, we applied the combination of SERS technique and machine leaning algorithms on glycogen analysis, which successfully predicted the origins of glycogen particles from a variety of organisms with convolutional neural network (CNN) algorithm plus attention mechanism having the best computational performance (5-fold cross validation accuracy = 96.97 %). In sum, this is the first study focusing on the discrimination of commercial glycogen particles originated from different organisms, which holds the application potential in quality control of glycogen-containing products.
polymer science,biochemistry & molecular biology,chemistry, applied