Nucleotide Sequence Classification of Paeonia Lactiflora Based on Feature Representation Learning

For the treatment of recurrent oral ulcer, total glucosides of paeony is an ideal drug. Long term application of total glucosides of paeony has low side effects and better patient compliance. Nucleotide sequence is of great significance in the field of Botany. In order to further study the pharmacology of Radix Paeonia Alba, the classification and prediction of nucleotide sequence of Radix Paeonia Alba is an important challenge. In this paper, we design a deep learning model based on graph neural network framework. First, each nucleotide sequence is extracted by k-mer algorithm, and then the extracted features are composed and put into the graph convolution layer for information transmission. Finally, the low-dimensional vector that integrates the whole sequence information can help us classify nucleotide sequences well. The final results of the experiment were 96.00% in Acc, 0.9387 in F1 score, 95.57% in Sn, 96.21% in Sp, and 0.9094 in MCC.