GANLDA: Graph attention network for lncRNA-disease associations prediction

Abstract Increasing studies have indicated that long non-coding RNAs (lncRNAs) play important roles in many physiological and pathological pathways. Identifying lncRNA-disease associations not only contributes to the understanding of biological processes, but also provides new strategies for the diagnosis and prevention of diseases. In this article, an end to end computational model based on graph attention network (GANLDA) is proposed to predict associations between lncRNAs and diseases. In our method, it combines heterogeneous data of lncRNA and disease as original features. Then, the principal component analysis (PCA) is used to reduce the noise of the original features. Further, the graph attention network is utilized to extract the useful information from features of lncRNA and disease. Finally, the multi-layer perceptron is employed to infer lncRNA-disease associations. The experimental results show GANLDA outperforms than other four state-of-the-art methods in 10-fold cross validation and devono test. The case studies also demonstrate that GANLDA is an effective method for lncRNA-disease associations identification.