EDLMFC: an ensemble deep learning framework with multi-scale features combination for ncRNA–protein interaction prediction

Abstract Background Non-coding RNA (ncRNA) and protein interactions play essential roles in various physiological pathological processes. The experimental methods used for predicting ncRNA–protein are time-consuming labor-intensive. Therefore, there is an increasing demand computational to accurately efficiently predict interactions. Results In this work, we presented ensemble deep learning-based method, EDLMFC, using the combination of multi-scale features, including primary sequence secondary structure tertiary features. Conjoint k-mer was extract protein/ncRNA integrating then fed into learning model, which combined convolutional neural network (CNN) learn dominating biological information with bi-directional long short-term memory (BLSTM) capture long-range dependencies among features identified by CNN. Compared other state-of-the-art under five-fold cross-validation, EDLMFC shows best performance accuracy 93.8%, 89.7%, 86.1% on RPI1807, NPInter v2.0, RPI488 datasets, respectively. results independent test demonstrated that can effectively potential from different organisms. Furtherly, also shown hub ncRNAs proteins networks Mus musculus successfully. Conclusions general, our proposed method improved interaction predictions anticipated providing some helpful guidance ncRNA functions research. source code datasets work available at https://github.com/JingjingWang-87/EDLMFC .