Comparison Of RNA Secondary Structure Based On Wavelet Transform And Fractal Dimension

With the further development and advancement of the Human Genome Project,the role of RNA molecules is being discovered increasingly.Since the outbreak of Newcastle pneumonia,RNA has gained significant attention as the genetic material of viruses.The study on the functions of RNA has been intensified,and because of the close relationship between its function and structure,the study of RNA's secondary structure has become a hot issue nowadays.Recently,there have been many methodologies for RNA secondary structure comparison,but the accuracy and generalizability of the algorithms need to be further improved.In this paper,two novel RNA secondary structure comparison methods are proposed for different types of RNA,which can obtain more accurate results.The main work is as follows.1.For the short RNA sequences,a comparison method for the secondary structure of short RNA sequences based on fractal dimension and wavelet transform is proposed.The triple vector curve representation of RNA is used to represent the RNA secondary structure as a signal graph and complete the feature extraction.Then,based on the multi-resolution property of Discrete Wavelet Transform(DWT),DWT was introduced to perform multivariate decomposition of the signal graph with biological information.The sliding window is applied and using the Higuchi algorithm to compute the fractal dimension of signal within a fixed-length window moving along the signal.Finally,the distance matrix is obtained by similarity calculation to construct the evolutionary tree.Several short RNA sequences from the commonly used RFAM database were selected and compared with existing RNA secondary structure comparison algorithms.The experiments show that the results obtained by the method in this paper are closer to the standard evolutionary tree with lower time complexity.2.In this paper,a novel method is proposed to compare the secondary structure of RNA for long non-coding RNA(lncRNA).The method is based on the TV-Curve representation of RNA secondary structure,introducing the discrete wavelet transform(DWT)and detrended fluctuation analysis(DFA)to obtain the fractal dimension,and analyzing the graphs obtained by feature extraction to generate distance matrices.The distance matrix was then clustered and visualized as an evolutionary tree.Using RNA virus and long non-coding RNA datasets for experiments,evolutionary trees are finally analyzed and compared.Results show that this method achieves more accurate results in the comparison of RNA secondary structures.In this paper,experiments are performed by using commonly used RNA databases and datasets.The results of the experiments demonstrate that the method proposed in this paper shows a significant improvement in classification accuracy for both short RNA sequence and long non-coding RNA.