Fast prediction of total sugar and protein content in Stropharia rugosoannulata by means of FTIR and chemometrics
Stropharia rugosoannulata is a new cultivated edible and medicinal mushroom favourably received by consumers in recent years.Its total sugar and protein content,is closely related to its nutritional values.However,the existent content detection methods are complex and time-consuming,and it is greatly significant to develop simple and rapid prediction methods.In this study,a total of 420 infrared spectra from fruiting bodies of S.rugosoannulata cultivated on seven different substrates was collected by Fourier transform infrared spectroscopy FTIR.The spectral pretreatment method was established by comparing the original spectrum and the pretreated spectrum,while the total sugar and protein characteristic spectra were selected by competitive adaptative reweighted sampling(CARS).Random forest(RF),partial least square regression(PLS),and support vector machine(SVM)were used for modeling.The results proved that the best prediction model for total sugar content was PLS,in which Rc was 0.992 8(the error was 0.007 2),RMSEC was 0.930 8,Rp was 0.981 4(the error was 0.018 6),RMSEP was 1.166 2,and RPD was 7.341 1;the best prediction model for protein content was RF,with Rc at 0.994 7(the error was 0.005 3),RMSEC at 0.380 3,Rp at 0.986(the error was 0.014),RMSEP at 0.749 1,and RPD at 8.437 5.The results showed that infrared spectroscopy combined with stoichiometry could quickly and accurately predict the total sugar and protein content of fruiting bodies of S.rugosoannulata.
Stropharia rugosoannulatainfrared spectrummodelingprediction model
NETL
NSTL
万方数据
