Background: MicroRNAs (miRNAs) are a kind of non-coding RNA, which plays an essential role in gene regulation by binding to messenger RNAs(mRNAs). Accurate and rapid identification of miRNA target genes is helpful to reveal the mechanism of transcriptome regulation, which is of great significance for the study of cancer and other diseases. Many bioinformatics methods have been proposed to solve this problem, but the previous research did not further study the encoding of the base sequence.
Results: In this study, we developed a novel method combining word embedding and deep learning for human miRNA targets at the site level prediction, which is inspired by the similarity between natural language and biological sequences. First, the wor2vec model was used to mine the distribution representation of miRNAs and mRNAs. Then, the data features are fully extracted automatically from temporal and spatial via the stacked Bidirectional Long short-term memory(BiLSTM) network. We compare the effects of different embedding methods on model accuracy in different deep learning models, and the results prove that using word2vec can improve the accuracy of deep learning models. In addition, we performed visual analysis on the distributed represented sequences and found hidden similarity relationships between bases. Finally, compared with different advanced methods and data sets, the results show that our proposed method has gotten better performance in multiple evaluation aspects.
Conclusions: We present a novel method for predicting miRNA target sites consisting of word2vec and the BiLSTM model and demonstrate that this method can realize automatic feature extraction and has higher accuracy. Furthermore, we process miRNA and mRNA as two languages for the first time and explore their biological significance through visual analysis.