High-precision Protein Sequence Comparison Based on Principal Component Analysis and Discrete Wavelet Transform

In the field of biomolecules, one of the most important way to predict protein function is to analyze proteins with similar sequences. The development of fast and accurate protein sequence comparison methods still faces more challenges. We put forward a method which use 9 main attributes of amino acids in this paper, and the low dimension attribute value is obtained by principal component analysis to simplify the calculation. The obtained first principal component is used to digitize protein sequence. Then calculate the similarity matrix using the approximate coefficients obtained by the discrete wavelet decomposition and fractal dimension. Finally, calculate the distance matrix and build a phylogenetic tree based on it. In the experimental part, a cluster analysis of 9 ND5 proteins was carried out and compared with existing models and MEGA software. The final results indicate the model that came up in this article is more effective.