RPI-HNBM: A hybrid network based on MLP-Mixer for ncRNA-protein interaction prediction

Protein-noncoding RNA (ncRNA) interactions play an important role in regulating gene expression and a variety of biological processes. However, current biological experimental methods for identifying ncRNA-protein interactions (ncRPIs) are costly and also time-consuming. When analyzing RNA sequences, their sequence lengths are challenging for convolutional neural network (CNN)-based methods, while existing methods simultaneously do not take into account the correlation between ncRNA and protein sequences. This paper proposes a hybrid network model based on structural features, RPI-HNBM (hybrid network based on MLP-Mixer), to predict ncRPIs using a Multi-Layer Perceptron Mixer (MLP-Mixer) network to address these issues. Firstly, ncRPIs are predicted by the Conjoint k-mer method for coding protein/ncRNA sequences. Then, they are inputted into a hybrid network combining the MLP-Mixer network to extract sequence features. Finally, the interactions between nucleotides and amino acids are modeled by the Attention module and output to the classification part for classification. In this paper, RPI-HNBM is evaluated on three publicly available datasets. The experimental results show that RPI-HNBM exhibits excellent performance on most of the evaluation metrics compared to other state-of-the-art prediction methods, which in turn demonstrates the effectiveness of the RPI-HNBM method for ncRPIs prediction.