Prediction of DNA methylation site status based on fusion deep learning algorithm

Abstract

DNA methylation is a crucial element of epigenetics and plays an important role in the evolution of life. As a result, detecting the status of DNA methylation becomes critically valuable. But since traditional biological experimental methods were unable to meet the actual needs, researchers began to employ machine learning and deep learning to aid biological experiments in determining methylation status. However, there are issues with feature acquisition, such as inconvenient extraction and high dimension. To address this issue, this paper proposes a feature extraction method based on convolution neural network (CNN) and recurrent neural network (RNN). Initially, the DNA methylation data used in this paper were obtained from the gene expression omnibus (GEO) database, and the data were preprocessed before use. Furthermore, we built a CNN and an RNN to extract features from DNA methylation data and then used feature splicing to find the best features. Eventually, we train the prediction model with a deep residual network and assess the model’s prediction performance with a confusion matrix. Compared with existing methods, we proposed method has better prediction performance.