Prediction Algorithm of DNA Sites Based on Weighted Feature Matrix.

DNA sites can realize various biological functions and play an important role in many biological activities. However, due to the complexity of feature extraction, feature learning is difficult. It is not conducive to realize high-precision identification and prediction of some DNA sites. In order to achieve efficient learning of DNA site features, we propose a DNA site prediction algorithm WFMP based on weighted feature matrix and polymerization strategy. The prediction algorithm learns the correlation between features and results, and assigns different weight information to each feature, emphasizing attention to important features and reducing the interference of invalid features. At the same time, WFMP effectively avoids the defects of a single algorithm by aggregating four traditional algorithms. Here, we apply the algorithm to classification and regression tasks respectively, which show high universality. The experimental results show that WFMP obtained an accuracy of 0. S966 in the classification task, achieving higher prediction accuracy. In the regression task, it obtained a root mean square error of 0.0051, showing an ultra-low error rate. In general, the WFMP algorithm is superior to the existing algorithms in all evaluation indicators, and has better portability and higher prediction accuracy. Based on the above studies, WFMP can not only detect hidden DNA binding sites, but also provide ideas for the study of efficient recombination systems.