Prediction of Retention Indices of Volatile Components in Rosmarinus officinalis L.Leaves by Neural Network Methodology
SHI Chunling
ZHOU Jun
GE Fengjuan
HE Changchun
ZHANG Tianshu
DU Xihua
Abstract：A neural network model for the quantitative structure-retention relationship of volatile components of Rosmarinus officinalis L.leaves was developed to rapidly predict their retention indices.Based on the molecular topology,a novel molecular structure index was defined with a strict discriminative effect on molecules.The connectivity index mX of the volatile components in Rosmarinus officinalis L.leaves was calculated.The relationship between the self-defined molecular structure characteristic indices,the software-calculated connectivity indexes and the retention index of the volatile components in Rosmarinus officinalis L.leaves was calculated.Eight indices including two kinds of self-defined molecular structure characteristic indices0C,1C,and six kinds of calculated molecular connectivity indices 0 X1X,2 X,4 X,5 X,and4Xpo were selected as the input layer variables,while the retention index was used as the output layer variable for the neural network.The neural network model was constructed using an 8-5-1 network structure.The overall correlation coefficient of the model was 0.994 2,and the average relative error of the predicted retention indices calculated by the model was 1.81％compared with the experimental values in the literature.There was a good non-linear relationship between the retention index,the newly defined molecular structure index and the molecular connectivity index.Group fragments such as—C—,—C=,—C＜,—O—,—OH,or=O were the main factors affecting the retention index of the volatile component in Rosmarinus officinalis L.leaves.
Keywords：Rosmarinus officinalis L.leafvolatile componentsretention indexmolecular structureneural network
Publication Date：2025-03-30
Online Publishing Date：2025-08-15(First online date of this platform, not the publication date of the document)
Pages：8( 79-86 )